Scalable Audio Feature Extraction -...
Transcript of Scalable Audio Feature Extraction -...
Scalable Audio Feature Extraction
Devon Leno Bryant
A Thesis Presented to the Faculty of
University of Colorado at Colorado Springs
in Candidacy for the Degree of
Master of Science in Computer Science
Adviser: Dr. Rory Lewis
April 2014
Recommended for Acceptance by
The Department of Computer Science
Dr. Rory Lewis Date
Dr. Jugal Kalita Date
Dr. Terrance Boult Date
Abstract
Feature extraction provides the foundation for many systems and research projects
dealing with machine learning. Extracting features from audio data can be compu-
tationally expensive and time consuming, particularly with larger sets of audio data.
This paper describes Safe, a system for scalable and efficient audio feature extrac-
tion in the Music Information Retrieval domain. The system builds upon previous
research in efficient audio feature extraction, taking advantage of both multicore pro-
cessing and distributed cluster computing to provide a scalable architecture for feature
extraction algorithms.
Lastly, the system’s performance is compared with current leading extraction li-
braries and a model for the system’s scalability is developed.
iv
Contents
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii
1 Introduction 1
2 Background and Related Work 3
3 System Description 6
3.1 Feature Representations . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.2 System Interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
4 Efficient and Scalable Extraction 11
4.1 Efficient Feature Extraction . . . . . . . . . . . . . . . . . . . . . . . 11
4.2 Parallel Feature Extraction . . . . . . . . . . . . . . . . . . . . . . . . 14
4.3 Distributed Feature Extraction . . . . . . . . . . . . . . . . . . . . . 17
4.3.1 Localized Data . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.3.2 Distributed Data . . . . . . . . . . . . . . . . . . . . . . . . . 18
5 Evaluation and Results 21
5.1 System Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
5.2 Dataflow Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
5.3 Scalability Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
v
5.3.1 Multi-Processor Scalability . . . . . . . . . . . . . . . . . . . . 31
5.3.2 Multi-Node Cluster Scalability . . . . . . . . . . . . . . . . . . 32
6 Conclusions 37
6.1 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
A Implemented Features 39
A.0.1 Frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
A.0.2 Window . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
A.0.3 Fast Fourier Transform . . . . . . . . . . . . . . . . . . . . . . 40
A.0.4 Power Spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . 41
A.0.5 Constant-Q Transform . . . . . . . . . . . . . . . . . . . . . . 41
A.0.6 Mel-Frequency Cepstral Coefficients . . . . . . . . . . . . . . . 42
A.0.7 Spectral Shape . . . . . . . . . . . . . . . . . . . . . . . . . . 43
A.0.8 Spectral Flux . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
A.0.9 Onset Detection . . . . . . . . . . . . . . . . . . . . . . . . . . 45
B Evaluation Dataset 47
References 115
vi
List of Tables
4.1 Plan level parallelism . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
4.2 Feature level parallelism . . . . . . . . . . . . . . . . . . . . . . . . . 15
5.1 Evaluation Dataset . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
5.2 Evaluation Dataset Genre Statistics . . . . . . . . . . . . . . . . . . . 22
5.3 Extraction Libraries and Versions . . . . . . . . . . . . . . . . . . . . 22
A.1 CQT Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
A.2 MFCC Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
A.3 Spectral Shape Parameters . . . . . . . . . . . . . . . . . . . . . . . . 44
A.4 Spectral Flux Parameters . . . . . . . . . . . . . . . . . . . . . . . . 45
A.5 Spectral Onset Parameters . . . . . . . . . . . . . . . . . . . . . . . . 46
B.1 Full Evaluation Dataset . . . . . . . . . . . . . . . . . . . . . . . . . 48
vii
List of Figures
3.1 Safe Web Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
4.1 Sequential dataflow for MFCC, CQT, and Spectral Shape features . . 12
4.2 Optimal dataflow extraction plan . . . . . . . . . . . . . . . . . . . . 12
4.3 Safe Distributed Deployment . . . . . . . . . . . . . . . . . . . . . . . 20
5.1 Feature extraction times (1332 songs, 5120 minutes of audio) . . . . . 25
5.2 Feature extraction times for increasing number of songs . . . . . . . . 26
5.3 Feature extraction times for single audio files of increasing length . . 27
5.4 Performance increase from dataflow analysis (single-threaded) . . . . 28
5.5 Percentage of total time taken by individual sub-feature actors . . . . 32
5.6 Feature extraction processor speedup . . . . . . . . . . . . . . . . . . 33
5.7 Feature extraction processor scalability . . . . . . . . . . . . . . . . . 34
5.8 Feature extraction cluster/node scalability (local audio files) . . . . . 35
5.9 Feature extraction cluster/node scalability (HDFS distributed audio
files) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
A.1 Sequential dataflow of the Constant-Q transform . . . . . . . . . . . . 42
A.2 Sequential dataflow for Mel-Frequency Cepstral Coefficients . . . . . 43
A.3 Sequential dataflow for Spectral Shape . . . . . . . . . . . . . . . . . 44
A.4 Sequential dataflow for Spectral Flux . . . . . . . . . . . . . . . . . . 45
A.5 Sequential dataflow for Spectral Onsets . . . . . . . . . . . . . . . . . 46
viii
Chapter 1
Introduction
The Music Information Retrieval (MIR) domain spans a wide variety of disciplines
and applications. MIR techniques are being leveraged by businesses and academics for
recommendation systems, classification, identification, automatic transcription, simi-
larity searching, and more. Despite the seemingly different nature of these problems,
many of these systems actually leverage similar audio feature extraction algorithms
at their core. These feature extraction algorithms can range from low-level digital
signal processing techniques to high-level machine learning algorithms. Most of these
algorithms are computationally intensive, both in resource usage and running times.
The amount of audio data and music being produced today is significant. Com-
panies such as Spotify are adding over 20,000 songs per day to their existing catalog
of over 20 million songs [39]. The current Apple iTunes catalog has over 26 million
songs [33]. Most of the existing MIR libraries and algorithms are not easily scalable
to work with larger datasets such as these.
Numerous libraries exist for audio feature extraction and many have focused on al-
gorithmic optimizations for efficient extraction, often using C++ for its performance
benefits in numerical computing. However, as modern computing continues to shift
towards more parallel architectures [3], it is clear that different programming models
1
and techniques are needed to take full advantage of these systems. Often, the im-
perative techniques used for optimal execution on a single processor are ill-suited or
difficult to transition to parallel environments [40].
Additional research has been done in extending existing libraries to work in dis-
tributed computing environments [28, 15, 5]. These systems are all based on a mas-
ter/worker division of nodes, which suffers from two main drawbacks. The primary
drawback is that a master node in any system provides a single point of failure, mean-
ing these systems do not address issues with fault-tolerance. The second drawback is
that a master node in a system often enforces bounds on the scalability; introducing
an upper bound when throughput is limited by the master node and forcing a lower
bound of two nodes from the master/worker separation.
This paper describes a high-performance concurrent system for audio feature ex-
traction based on modern techniques in parallel computing and reactive program-
ming. Scalability of the system is fault-tolerant and addresses both vertical scalability
through multicore processing and horizontal scalability through distributed cluster-
ing. While the primary focus of this research is on performance and scalability, there
is a strong secondary goal towards providing a general extensible MIR framework for
others to re-use and build upon.
2
Chapter 2
Background and Related Work
There have been many MIR feature extraction libraries developed over the last decade,
each with different feature sets and focus. This section outlines some of the most
popular and relevant libraries in audio feature extraction, though many more exist.
One of the earliest libraries to provide audio feature extraction that is still used
today is Marsyas [41]. Marsyas is an efficient C++ framework for general dataflow
audio processing that has been used for various analysis and synthesis tasks. Marsyas
provides both a command-line interface and a graphical editor for building data flows.
Marsyas has also been extended to perform batch extraction in distributed environ-
ments [5]. The distributed model relies on a custom dispatcher node sending audio
clips over TCP to known worker nodes for feature extraction.
Yaafe [27] is a fast and efficient library for audio feature extraction written mostly
in C++ with a Python interface. The focus of Yaafe is on efficiency, utilizing dataflow
graphs to eliminate redundant calculations between different features. Yaafe is mostly
a command line tool and supports output in CSV or HDF5 formats.
jAudio [30] is a Java-based audio extraction framework developed primarily for
MIR researchers. The framework provides a GUI for selecting features and filling out
parameters. Results can be exported to XML or Weka’s ARFF format for further
3
experimentation or machine learning. jAudio also provides a mechanism for describ-
ing dependencies between features, similar to Yaafe, for the purpose of eliminating
duplicate calculations.
OMEN [28] is a distributed extraction framework that leverages jAudio as its base
extraction engine. The distribution model for OMEN is based on a hierarchy of differ-
ent node types: a master node provides the overall coordination, library nodes store
collections of audio data, and worker nodes perform the actual feature extraction. All
communication in OMEN is handled through SOAP-based web services.
Sonic Annotator [8] is a command-line tool for running batch feature extraction
on large numbers of audio files. Sonic Annotator is based on VAMP [19] plugins, a
general framework and C++ API for audio feature extraction algorithms. Numerous
VAMP plugin libraries have been developed by different research groups and are
available for use. Sonic Visualiser, a complimentary tool to Sonic Annotator, provides
a helpful user interface for feature exploration and many different graphical displays
for visualizing features. Unfortunately, algorithms and results in VAMP cannot be
shared between plugins which leads to some inefficiencies during extraction.
iSoundMW [15] is a distributed feature extraction framework and similarity search
engine. The distribution model uses a single master node that sends small segmented
frames of audio data to registered client worker nodes for feature extraction.
MIRtoolbox [26] is a Matlab-based MIR feature extraction library that takes ad-
vantage of existing visualizations and algorithms available in the Matlab ecosystem.
MIRtoolbox provides some explicit mechanisms for composing features in stages from
lower-level or intermediary calculations. However, memory management is an issue
since the entire file and all feature calculations are held in memory. MIRtoolbox pro-
vides some parallel extraction capabilities, allowing concurrent extraction of separate
files.
4
openSMILE [17] is a C++ based library focused on real-time Speech and MIR fea-
ture extraction. Batch extraction capabilities are provided as well as multi-threaded
support for parallel extraction on a single machine. openSMILE is based on a com-
plex shared data memory system which allows single-component writes and multi-
component reads of matrix data and leverages ring-buffers to manage memory. Dif-
ferent output formats are provided including CSV, ARFF, LibSVM, and HTK.
5
Chapter 3
System Description
3.1 Feature Representations
The primary abstraction for any extraction library is the description of a Feature.
Within the Safe system, each feature is described by a set of required and optional
parameters needed to perform its calculations. Features in audio processing are often
built on top of other features or share common steps in their calculations. For exam-
ple, the Fast Fourier Transform (FFT) is generally a shared step for any algorithms
working in the frequency domain. For that reason, each feature also describes its own
dataflow as a sequence of smaller computational steps. This dataflow representation
provides two major benefits. The first benefit is algorithm reuse; new algorithms can
leverage existing code for common calculations. The second benefit is that duplicate
calculations between features can often be eliminated through simple dataflow analy-
sis on the provided feature sets. Section 4.1 describes this dataflow analysis technique
in more detail. For a full list of implemented features, parameter descriptions, and
dataflow sequences see Appendix A.
Externally, features are described through a simple syntax similar to the syntax
used by Yaafe [27]. Listing 1 shows a simplified Backus-Naur Form (BNF) of the
6
feature extraction syntax used by the system and listing 2 shows an example entry of
this syntax. Each entry contains three primary sections:
1. name - A unique name for the feature extraction.
2. feature - The feature to be extracted and any feature parameters such as the
step size, window type, etc.
3. output - Optional output parameters such as the directory to write to or the
precision of the results.
Listing 1 Simplified Feature Syntax BNF
〈identifier〉 |= { A . . .Z | a . . . z | 0 . . . 9 }〈name〉 |= 〈identifier〉〈feature〉 |= 〈identifier〉
〈parameter〉 |= 〈identifier〉 ”=” ( 〈quoted string〉 | 〈number〉 )
〈parameters〉 |= 〈parameter〉 { ”,” 〈parameter〉 }〈feature definition〉 |= 〈name〉 ”:” 〈feature〉 [ 〈parameters〉 ] [ ”-out” 〈parameters〉 ]
Listing 2 MFCC Feature Extraction Description
mfcc: MFCC frameSize=256, stepSize=256, windowType=”hamming”
3.2 System Interfaces
Within the system, there are two primary interfaces for running feature extraction
over existing datasets. The first interface provides command-line execution for run-
ning bulk extraction on a single machine. Listing 3 outlines the syntax and parameters
used in the command-line interface.
The two required parameters for any extraction are:
7
Listing 3 Command Line Syntax
− i < f i l e > | −−input < f i l e >input path ( f i l e or d i r e c t o r y ) f o r audio f i l e ( s ) to p roce s s
−r | −−r e c u r s i v ef l a g to proce s s audio data in sub−d i r e c t o r i e s ( d e f a u l t = f a l s e )
−p < f i l e > | −−plan < f i l e >path to f e a t u r e e x t r a c t i o n plan f i l e
−f <value> | −−f e a t u r e <value>d e s c r i p t i o n and parameters f o r a s i n g l e f e a t u r e to e x t r a c t
−s <value> | −−sample−r a t e <value>sample ra t e in Hz o f a l l audio f i l e s ( d e f a u l t = 44100)
−o < f i l e > | −−output−d i r < f i l e >d i r e c t o r y to wr i t e f e a t r e output to ( d e f a u l t = ’ . / ’ )
−m | −−metr i c sf l a g to capture / p r i n t metr i c s ( d e f a u l t = f a l s e )
1. Input - The individual audio file or directory containing audio files to run ex-
traction on.
2. Plan File or Feature - A file containing the feature set and parameters or an
inline description of the feature and its parameters.
The remaining command-line parameters are optional and use the specified de-
faults if no input is provided. One particular parameter of interest is the sample
rate. Safe currently requires all audio data to be encoded with the same sample rate.
This limitation exists because several features use the expected sample rate to pre-
compute values for use in their calculations. For example, the Mel-Frequency Cepstral
Coefficients (MFCC) algorithm uses the sample rate along with other parameters to
pre-compute and cache its mel filter bank. In cases where a dataset contains audio
with different sample rates, the audio data needs to be pre-processed and re-sampled
into a common rate.
The Safe system also has the ability to capture and print out metrics for a given
extraction run. If enabled, Safe will capture the following metrics for every sub-feature
computation:
8
• Total Messages - The total number of times the sub-feature calculation was
called during extraction.
• Total Time - The accumulated total time spent calculating the sub-feature.
• Min, Max, Median, Mean Times - The minimum, maximum, median, and mean
times spent in a single calculation of the sub-feature.
The second system interface, used primarily for controlling extraction in dis-
tributed environments, is a web-based user interface. The web interface uses REST
[18] communication for executing extraction and monitoring the progress and system
status. Each extraction request is assigned a unique id that can be used for querying
the status of the extraction. There are three possible responses for extraction status
queries:
1. Ongoing - Specifies the number of audio files that have been processed along
with the total number of files.
2. Failed - Gives a description of any failures that occurred during extraction.
3. Finished - Specifies the total number of files processed and the time taken to
run the extraction (in milliseconds).
Figure 3.2 below shows a screenshot of the web interface for the system. This
example screenshot shows the three different status results for extraction requests.
9
Chapter 4
Efficient and Scalable Extraction
The primary goal of this research is to provide an efficient framework for audio feature
extraction that can easily scale in both vertical and horizontal directions. This is
achieved by extending the techniques and algorithms used in other MIR libraries
and adapting them to work with functional, asynchronous and reactive programming
models. Safe is implemented in Scala [25] and leverages the Akka [2] library for
parallel and distributed computing.
4.1 Efficient Feature Extraction
As mentioned in chapter 3, features can often be described by dataflow sequences of
smaller computational steps or sub-features. Figure 4.1 shows an example of the step
sequences for calculating three different features: Mel-Frequency Cepstral Coefficients
(MFCC) [12], Constant-Q Transform (CQT) [6], and Spectral Shape [21].
By breaking down features into smaller steps, efficiencies can be gained by elim-
inating shared calculations. A set of multiple features in the system is combined
to form a Feature Plan. When shared calculations between feature sequences are
eliminated in a feature extraction plan, the plan naturally forms a directed acyclic
dataflow graph. This is a common technique in dataflow analysis that is used by other
11
MIR feature extraction libraries [29, 27] to eliminate redundant calculations. This
technique is similar to the problem of Common Subexpression Elimination [10] in
compiler optimization. Figure 4.2 below shows an example of the optimized dataflow
extraction plan for the same three features described earlier.
Audio in Frame Window FFT Mag MFCC
Audio in Frame Window FFT CQT
Audio in Frame Window FFT MagSpectralShape
Figure 4.1: Sequential dataflow for MFCC, CQT, and Spectral Shape features
CQT
FFT
Mag
MFCC
SpectralShape
WindowFrameAudio in
Figure 4.2: Optimal dataflow extraction plan
Feature extraction plans in Safe are encoded using adjacency lists. Each plan holds
a single feature or sub-feature computation and zero or more downstream plans to
pass results to. Algorithm 1 shows the basic data representation for feature extraction
12
plans. Algorithm 2 shows the heuristic algorithm that is used to combine multiple
features or plans into optimal dataflow graphs.
Algorithm 1 Feature Extraction Plan Representation
case class Plan(feature: Feature, children: Seq[Plan])
Algorithm 2 Dataflow Graph Optimization
def dataflow(plans: Seq[Plan]): Seq[Plan] =
plans.foldLeft(Nil)((merged, plan) => merge(merged, Seq(plan)))
def merge(as: Seq[Plan], bs: Seq[Plan]): Seq[Plan] = {
val merged = for {
a <- as
b <- bs if (a.feature == b.feature)
} yield Plan(a.feature, merge(a.children, b.children))
(as ++ bs).filterNot(
p => merged.exists(_.feature == p.feature)) ++ merged
}
13
4.2 Parallel Feature Extraction
With the processing model expressed as a directed acyclic dataflow graph, audio data
can be streamed through the pipeline to extract the various features. This approach
lends itself to numerous levels of parallelism:
• File Level - Extraction can be run concurrently on multiple files.
• Plan Level - Any time a plan branches into separate paths, each path can run
concurrently on the results of the parent calculation.
• Feature Level - When audio data is split into separate chunks (frames), each
stage of a feature extraction sequence can concurrently operate on a different
section of the data.
File level parallelism is a fairly self-explanatory concept and is the approach used
in [26]. It is the most obvious form of parallelism and requires little to no coordination
between processes. Tables 4.1 and 4.2 show examples of plan level and feature level
parallelism where each row represents a different processor and each column repre-
sents possible computations over time. With plan level parallelism, different steps can
run concurrent calculations on the same data outputs of a parent step. For example,
table 4.1 shows that the CQT and Magnitude calculations can run concurrently on
the output of the FFT. With feature level parallelism, a given calculation may be
executing on a section of audio data concurrently with downstream calculations run-
ning off previous results. In table 4.2, the example highlights that a Window function
can pass its output to the FFT function and immediately begin working on the next
frame. Note: These tables are used for illustrative purposes of where parallelism can
occur and do not represent the actual thread scheduling and division within the sys-
tem. In the actual running system, some steps will take much longer than others and
a single thread may finish multiple steps in the time another thread takes to execute
14
one. Additionally, different combinations of file, plan, and feature level concurrency
may occur together.
FFT(window1) Magnitude(fft1) MFCC(mag1)CQT(fft1) Spectral Shape(mag1)
Table 4.1: Plan level parallelism
frame1 Window(frame1) FFT(window1) CQT(fft1)frame2 Window(frame2) FFT(window2)
frame3 Window(frame3)frame4
Table 4.2: Feature level parallelism
Managing these different levels of concurrency with traditional thread scheduling
and lock management would be complicated and difficult. Instead, an Actor model
[1] for concurrency is used.
Actors are objects that encapsulate state and behavior and communicate exclu-
sively through messaging. Messaging between actors is never direct. Instead, com-
munication occurs through the use of recipient mailboxes. Each actor has exactly one
mailbox and messages are dequeued and processed by the actor in a serial fashion.
This serial processing guarantee helps eliminate synchronization and locking concerns
within the actor’s actual processing logic. Conceptually, actors can be thought of as
having their own light-weight isolated threads. It is the responsibility of the actor
container to manage the actual scheduling and execution against real threads.
Actors also provide a rich hierarchy and supervision system. Each actor can create
any number of child actors to execute sub-tasks. When an actor creates another actor,
it becomes a supervisor and is responsible for delegation and control of the child actor.
When a child actor fails for any reason, the supervisor (parent) is notified of the failure
and is responsible for handling the failure. When a subordinate fails, the supervising
actor has the option to resume the subordinate (keeping any existing state), restart the
15
subordinate (clearing out any previous state), terminate the subordinate, or escalate
the failure (failing itself).
Actor models, being asynchronous and message-based, provide a natural encoding
for stream based dataflow processing. Within Safe, each computational step in the
dataflow graph is represented by an actor and the intermediary results (messages)
between actors are immutable. By encoding the feature extraction dataflow graphs
as actor hierarchies in this fashion, all three levels of parallelism mentioned (file, plan,
and feature) are provided with no additional components required. The dynamic
nature of Akka’s actor system make it simple to create complex fault-tolerant feature
extraction hierarchies. Within a feature extraction plan, each actor usually takes on
one of four common messaging roles described in [24]:
• Splitter - Actors that take a single message or result and split the work into
smaller pieces for downstream processing. Splitters provide much of the par-
allelism in feature extraction. File level parallelism is provided by splitting a
message with multiple file locations (or a directory) into individual messages
for each file. Feature level parallelism is provided by a Frame actor, splitting
an individual file into smaller chunks (frames) to perform further computations
on.
• Transform - The most common actor type in a feature extraction plan is a
transform actor. These actors provide simple mapping functions from one type
to another and send the result on for further processing. Examples of transforms
are Windowing functions, FFT, Power Spectrum, MFCC, Spectral Shape, etc.
• Resequencer - Actors that take a stream of potentially unordered messages and
send them out in an ordered sequence. Sequencing is important when writing
out frame-based extracted features. For example, when writing out to a CSV
16
file the features need to be in the same order as the frames in the source audio
file.
• Aggregator - Aggregators are used to calculate features that go across multiple
frames in an audio file or across an entire dataset. Some example aggregate
functions are Spectral Flux or any max/min/average/total/etc. calculation
across audio files.
By default, a single actor is created for each node in the dataflow graph and the
actor messaging hierarchy mirrors the arcs in the graph. For many systems, the con-
currency provided by the file, plan, and feature level divisions is sufficient. However,
for highly concurrent systems (systems with a large number of processors), the Split-
ter and Transform actors can be replaced with actor pools and a round-robin routing
strategy. This strategy allows concurrent execution of the same computational step on
different frames. For example, multiple FFT calculations could be run concurrently
on different frames. In the current system, Resequencers and Aggregators cannot be
placed into pools because they need to maintain a temporary state across multiple
messages. Because of this requirement, scalability is often limited by features that
require aggregate calculations.
4.3 Distributed Feature Extraction
The distribution model for the system leverages Akka’s clustering framework [9].
Akka’s cluster system provides a fault-tolerant decentralized membership service
based on Amazon’s Dynamo [14] store. The cluster architecture uses gossip pro-
tocols [42] to manage node membership within the cluster. Using the gossip protocol,
the current state of the cluster is gossiped between nodes and preference is given to
members that have not yet seen the latest version. Initiation into a cluster is handled
by a node sending a join message to one of the existing nodes in the cluster. In order
17
to provide a fault-tolerant solution, failures in the cluster need to be detected and
communicated to the rest of the system. For this, Akka uses a Phi Accrual Failure
Detector [23].
While previous research in distributed audio feature extraction exists [28, 15, 5],
they are all based on a master/worker division of nodes. By leveraging a decentralized
and masterless clustering framework, there is no single point of failure or single point
bottlenecks in the system. Since the clustering framework is Actor based as well, it
is simple to scale the feature extraction framework from a single machine to multiple
nodes.
4.3.1 Localized Data
The simplest form of distributed feature extraction assumes that a collection of audio
files is already distributed across the cluster onto the local file systems. This config-
uration is similar to the library node concept used by OMEN [28]. From there, bulk
extraction can be performed across the cluster through a simple broadcast message
to all nodes in the cluster. While this provides a simple and scalable solution, this
approach is not fault-tolerant or flexible. This distribution model couples the extrac-
tion execution to the machines containing the data. If a node is down or fails, there
is no way to run extraction over the audio data stored on that node.
In this case, the time taken to extract all features will be limited by the longest
running node in the cluster. For optimal extraction times, the data needs to be evenly
balanced across the cluster and nodes within the cluster should be homogeneous, i.e.
each machine should have an equal number of computing resources.
4.3.2 Distributed Data
A more robust and fault-tolerant model for distributed extraction requires a dis-
tributed file system such as Hadoop HDFS [38]. Hadoop provides a whole suite of
18
capabilities for distributed analysis and transformation of data using the MapReduce
[13] paradigm. However, Safe’s cluster extraction only leverages the distributed file
system from Hadoop.
In HDFS, there are two types of nodes: NameNodes and DataNodes. NameN-
odes provide the overall management and coordination of directory hierarchies and
location of the data within HDFS. File content is split into chunks, typically 128 mb
in size, and those chunks are stored on the local file systems of DataNodes. Fault-
tolerance in HDFS is provided through replication. Individual chunks are replicated
and distributed onto multiple DataNodes. In a standard configuration NameNodes in
a hadoop cluster provide a single point of failure, though a High Availability (HA) fea-
ture is provided that allows redundant NameNodes to be configured for fast failover.
HDFS clients typically access files through a TCP protocol, but a short-circuiting
mechanism [34] is provided for cases when the data and processing are co-located on
the same machine.
Figure 4.3.2 shows the distributed deployment architecture of the system using
HDFS. Since the Safe cluster is masterless, any number of nodes in the cluster can
be configured to serve http requests. Safe nodes can be co-located with HDFS nodes
or run on separate clusters.
When an extraction request is made to one of the nodes in the Safe cluster, the
receiving node will query the NameNode for all the paths of the audio data and route
extraction of individual audio files to different nodes in the cluster. The primary
routing logic uses a Round-Robin routing strategy to distribute the extraction work-
load across the cluster. Some experimentation with routing strategies that take data
locality into account was also performed. In this case, the routing logic favors sending
extraction requests to the same machine where the data is located, taking advantage
of the HDFS short-circuit reads. When different nodes in a cluster are unbalanced,
i.e. some nodes have a larger heap, more CPU, etc., an adaptive load balancing router
19
Figure 4.3: Safe Distributed Deployment
should be used instead to distribute feature extraction. Akka provides a probabilistic
router based on weighting from the following metrics:
• Heap - Remaining JVM heap capacity.
• Load - System load average over the past minute.
• CPU - Percentage of CPU utilization.
20
Chapter 5
Evaluation and Results
5.1 System Evaluation
In order to evaluate the system’s performance and scalability, a large dataset of music
audio was required. The dataset used for testing was a private collection of 1,332
songs, approximately 5,120 minutes of audio. All files were encoded in CD-quality
44.1 KHz 16-bit stereo WAV format. The overall dataset was split into smaller subsets
to allow for testing on varying sized datasets. Tables 5.1 and 5.2 show some general
statistics about the datasets used in testing. For a full listing of all the songs, see
Appendix B.
Safe was compared with three other feature extraction libraries for performance:
Yaafe, openSMILE, and Sonic Annotator. Yaafe and openSMILE in particular were
chosen because of their strong performance characteristics. Table 5.3 shows the ver-
sions of each library used for this evaluation. Initial evaluations also included MIR-
Toolbox, though its performance was significantly worse and the results have been
left out of these benchmarks.
21
Audio Files Total (minutes) Max (minutes) Min (minutes) Avg (minutes)1 2.5 2.5 2.5 2.52 5.0 2.5 2.5 2.54 10 3.49 1.51 2.59 20 4.04 1.09 2.2217 40 4.56 0.92 2.3535 80 4.56 0.59 2.2961 160 5.91 0.59 2.62107 320 6.98 0.41 2.99200 640 11.35 0.41 3.2371 1280 11.35 0.41 3.45707 2560 15.72 0.23 3.621332 5120 32.25 0.11 3.84
Table 5.1: Evaluation Dataset
Genre Song CountProgressive Metal 19
Folk Rock 2Pop Rock 3New Wave 32Indie Rock 223
Metal 79Jazz 171Punk 386Swing 22
Stoner Metal 27Alt-Country 10
Folk 30Rockabilly 3
Rock 31Alternative 213
Blues 6Hip-Hop 74
Table 5.2: Evaluation Dataset Genre Statistics
Library VersionYaafe 0.64openSMILE 2.0-rc1Sonic Annotator 1.0Safe 0.1
Table 5.3: Extraction Libraries and Versions
22
The benchmark used for comparing the different libraries was a measure of the
total time taken to extract six different features and output the results to CSV files.
The six features picked for the benchmark were:
• Mel-Frequency Cepstral Coefficients (13 coefficients, 40 filters)
• Spectral Flux
• Spectral Centroid
• Spectral Spread
• Spectral Skewness
• Spectral Kurtosis
This particular feature set was chosen because each library has support for these
features and the algorithms are fairly standard and not likely to vary between imple-
mentations. For all features, the frame size was 1024 samples and the step (hop) size
was 512. All of the scripts and configuration used for this evaluation are available
online [37]. In the interest of fairness, attempts were made to provide configurations
for optimal performance with each library.
Yaafe was compiled with FFTW [20] for fast FFT calculations. Since Yaafe per-
forms dataflow analysis to optimize execution, no other special configuration was
required. Since openSMILE has support for multi-threaded concurrency, experimen-
tation was performed to find an optimal number of threads for the test. Ultimately,
openSMILE was configured to run with a single thread because all tests with multi-
threaded configurations appeared to degrade the performance. For openSMILE, the
dataflow graph is encoded explicitly into the configuration since there is no dataflow
analysis support. For Sonic Annotator, the LibXtract [7] plugins were used to calcu-
late the spectral shape features (centroid, spread, skewness, and curtosis), while the
23
BBC plugins [11] and Queen Mary University plugins [32] respectively were used to
calculate the Spectral Flux and MFCC features. Sonic Annotator does not provide
any support for dataflow optimizations, so many of the algorithms are running re-
dundant calculations. For this benchmark Safe was configured using Akka’s default
Fork-Join executor, which calculates its thread pool size based on the number of
processors p and a parallelism factor ν (default 3.0):
poolsize = dpνe (5.1)
All evaluations were run on a Macbook Pro, OS X 10.7.5, with a 2.5 GHz Intel
Core i7 processor 8 GB 1333 MHz DDR3 RAM. For each evaluation, the mean time
and standard deviation were calculated from ten iterations. The standard deviation
for each library was relatively small. In many of the plots, the standard deviation is
left out because the bars are indistinguishable from the plotted points.
Figure 5.1 shows the overall time taken by each library to run extraction over the
full dataset. Figure 5.2 shows a plot of each libraries performance as the number
of songs increases. Yaafe, openSMILE, and Safe (single-threaded) are all very close
in terms of performance. In this benchmark, the multi-threaded version of Safe is
roughly four times faster than the other solutions.
24
Sonic Annotator Yaafe openSMILE Safe-1 Safe
20
40
60
80
100
120112.39
89.24 88.09
79.81
23.56
Extr
acti
onT
ime
(min
ute
s)
Figure 5.1: Feature extraction times (1332 songs, 5120 minutes of audio)
In order to verify that the parallelism in Safe is not limited to concurrent extraction
of multiple files, a benchmark was setup to measure extraction times on single audio
files of increasing length. Figure 5.3 shows the results of this benchmark. For this
benchmark, Safe was configured to use Akka’s fixed thread pool dispatcher with five
threads. These results indicate that the Safe system has a more significant fixed
startup cost than the other libraries but better variable performance with increasing
recording lengths.
25
101 102 103 104
10−1
100
101
102
Minutes of Audio
Extr
acti
onT
ime
(min
ute
s)
openSMILEYaafe
Sonic AnnotatorSafe
Figure 5.2: Feature extraction times for increasing number of songs
26
1 2 3 4 5 6 7 8 9 100
2
4
6
8
10
12
14
Minutes of Audio
Extr
acti
onT
ime
(sec
onds)
openSMILEYaafe
Sonic AnnotatorSafe
Figure 5.3: Feature extraction times for single audio files of increasing length
27
5.2 Dataflow Analysis
In order to test the benefits of dataflow analysis, Safe’s performance was measured
both with and without dataflow optimizations. Testing was performed using the
same dataset and feature set described previously in section 5.1. For the sequential
case, each audio file was loaded only once but the individual extraction algorithms
were run independently with no shared calculations. This setup is very similar to the
way many feature extraction libraries (including Sonic Annotator) work. Figure 5.4
shows the extraction times both with dataflow optimizations and without (sequential).
For this particular feature set, we see a 33% decrease in the overall extraction time
when dataflow optimizations are applied. However, the benefits can vary significantly
depending on the selected feature sets and input parameters.
0 20 40 60 80 100 120 140 160
0
20
40
60
80
100
120
140
160
180
200
220
Minutes of Audio
Extr
acti
onT
ime
(sec
onds)
SequentialDataflow
Figure 5.4: Performance increase from dataflow analysis (single-threaded)
28
5.3 Scalability Analysis
The techniques used to measure and describe scalability of the system are all based
on Gunther’s Universal Scalability Law (USL) [22]. Within this model, one of the
primary measurements for parallelism in a system is the system’s speedup factor.
Speedup is defined as the ratio of elapsed time on a single processor T1 to the elapsed
time on p processors Tp.
S(p) =T1Tp
(5.2)
Speedup measurements are taken by executing a fixed workload on increasing
numbers of processors (or machines). The result is a quantification of the reduced
execution time for adding more processors to the system. Another important measure-
ment of a system’s parallelism, based on speedup, is efficiency. Efficiency measures
the average speedup per processor, which is defined as:
E(p) =S(p)
p(5.3)
In theory, a system that is perfectly parallel will have linear speedup T1/p. In
practice, however, there are other factors that make linear speedup in systems un-
realistic. The Universal Scalability Model was developed to take these additional
factors into account and give a realistic description of a systems true scalability. The
two primary factors involved in this model are:
• Contention (σ) - The serial fraction of the total execution. This represents the
portion of an algorithm that cannot be split up and needs to run serially.
• Coherency (κ) - The additional overhead of managing extra processors and
interprocessor communication.
29
Taking these factors into account, the Universal Scalability equation for system
capacity C(p) is defined as:
C(p) =p
1 + σ(p− 1) + κp(p− 1)(5.4)
Equation 5.4 describes a concave function. Under this model, most systems will
have a maximum capacity at p∗. Adding more processors beyond p∗ will only degrade
performance. The maximum capacity can be obtained through the following equation:
p∗ =
⌊√1− σκ
⌋(5.5)
The σ and κ parameters in the USL are calculated by performing regression anal-
ysis on the measured system throughput X(p). The full procedure for calculating
these parameters is outlined in [22].
All scalability testing of the system was performed on the University of Colorado
at Colorado Springs virtual cloud infrastructure [31] running a VMWare vSphere
5 hypervisor. The particular cluster used consisted of 8 physical HP machines w/
2.4GHz Intel Xeon E5530 CPUs. The total available resources in the cluster consisted
of:
• 64 Processor Cores
• 153 GHz total CPU
• 512 GB Memory
• 55 TB Connected Storage
All virtual machines used in testing were running a 64-bit CentOS 6.4 operating
system.
30
5.3.1 Multi-Processor Scalability
In addition to providing a simple model for developing concurrent applications, the
Actor model provides a simple theoretical foundation for estimating scalability. Since
each actor processes messages from its inbox serially, the contention factor σ in an
actor system can be estimated by running the system with a single processor (or
single-threaded) and measuring the percentage of time spent in each actor. The
largest percentage of time spent in a single actor should be very close to the overall
contention factor.
Using the same features and dataset described in section 5.1, figure 5.5 below shows
the percentage of time spent computing each step (sub-feature) for different pool sizes.
In the case where a single actor is dedicated to each sub-feature computation (pool
size of 1), reading in the input audio and creating the frames takes approximately 21%
of the time on a single processor. Therefore, the theoretical serial contention factor
σ1 should be ≈ 0.21. In the current system, features that require aggregation (such as
Spectral Flux) cannot be pooled and only a single actor is created regardless of pool
size. Therefore, these features will often become the dominant factor in scalability
when pooling is used. With this feature set, the Spectral Flux calculation takes
approximately 13% of the time on a single processor. For pool sizes of two or more,
the contention factor σn should be ≈ 0.13.
In order to test the system’s actual multi-processor scalability, the same extraction
test was run on virtual machines with an exponentially increasing number of single-
core processors p = 1, 2, 4, 8, . . . , 64. Multiple tests were run to capture the impact
of increasing actor pool sizes. Figure 5.6 shows the measured system speedup. As
expected from the theoretical model, no performance increases were seen for pool
sizes greater than two.
Figure 5.7 shows the measured throughput and modeled curve fits based on the
USL calculations. From these equation parameters, our maximum capacity p∗ for
31
Fr (I/O) FFT MFCC Shape Flux CSV (I/O)
2
4
6
8
10
12
14
16
18
20
22 21
16
12
10
13
6
12
87
5
13
6
9
5 54
13
6
%O
fE
xtr
acti
onT
ime
Pool 1 Pool 2 Pool 3
Figure 5.5: Percentage of total time taken by individual sub-feature actors
this particular feature set is 20 processors for a pool size of one (single actor per
sub-feature computation) and 28 processors for a pool size of two. The calculated
contention factors σ1 = 0.203 and σ2 = 0.123 were very close to the theoretical
expected values of 0.21 and 0.13.
5.3.2 Multi-Node Cluster Scalability
The same USL model used to evaluate Safe’s multi-processor scalability was used
for measuring the distributed cluster scalability. Instead of increasing the number of
processors on a single virtual machine, tests were run on increasing numbers of virtual
machines. Each node (VM) in the cluster was given the same number of resources
32
0 10 20 30 40 50 60 70
1
1.5
2
2.5
3
3.5
4
4.5
5
5.5
Number of Processors
Sp
eedup
Fac
torS
(p)
Pool Size 1Pool Size 2
Figure 5.6: Feature extraction processor speedup
(processors, disk space, etc.). In most hosted VM environments, multiple VM’s will
often share and compete for physical resources such as disk space and processors. In
an attempt to minimize this contention for these tests, each VM was mapped to a
separate physical hard drive and processor. Since the physical cluster only provided
six separate disk partitions and eight separate processors, the maximum number of
nodes tested was six.
The first test, as outlined in section 4.3.1, relied on the dataset being distributed
evenly across the cluster onto the local drives. For this test, the audio files were
distributed across the cluster and placed in the same directory location on each node.
Communication between nodes in this configuration is minimal and only relies on
the initial broadcast message to perform extraction and the final aggregation step
determining when all nodes have finished.
33
0 10 20 30 40 50 60 70
40
60
80
100
120
140
160
180
200
σ1 = 0.203, κ1 = 0.001949
σ2 = 0.123, κ2 = 0.001101
Number of Processors
Extr
acti
onT
hro
ugh
put
(sec
ofau
dio
/se
c)
Pool 1 (measured)Pool 1 (USL model)Pool 2 (measured)
Pool 2 (USL model)
Figure 5.7: Feature extraction processor scalability
Figure 5.8 shows the measured and USL calculated scalability of this model. As
expected, because of the minimal coordination between nodes, the scalability is close
to linear. Based on the calculated contention and coherency, the maximum capacity
p∗ of this configuration is 82 nodes.
While the previous localized data configuration has close to ideal scalability, it is
a poor choice for fault-tolerance or flexibility in the distribution model. The second
test, as described in 4.3.2, relied on the audio data being distributed through HDFS.
The HDFS configuration for these tests specified a replication factor of 3 and a block
size of 128 MB. For CD quality wav files (44.1 KHz, 16-bit stereo), 1 minute of
audio is approximately equal to 10 MB. Therefore, most songs under 12 minutes in
length were stored in a single block. Each node ran an instance of the Safe software
34
1 2 3 4 5 61
2
3
4
5
6
7
8
9
10
σ = 0.0113, κ = 0.000144
Number of Nodes
Extr
acti
onT
hro
ugh
put
(min
ofau
dio
/se
c)
MeasuredUSL model
Linear
Figure 5.8: Feature extraction cluster/node scalability (local audio files)
and doubled as an HDFS data node. With each new node added, the cluster was
re-balanced to more evenly distribute the audio files.
Initial testing leveraged a cluster-aware router that attempted to route extraction
of individual files based on data locality. In this case, preference was given to nodes
where the physical audio files and processing were co-located in an attempt to take
advantage of HDFS short-circuit local reads. However, testing revealed that the
HDFS re-balancing algorithm did not provide a truly even distribution of the data,
causing the distributed workload to become unbalanced. Instead, a standard round-
robin routing strategy was used which gave better overall performance.
Figure 5.9 below shows the measured and USL calculated scalability of the HDFS-
based distribution. Not surprisingly, the contention and coherency factors are slightly
higher in this configuration, but the fault-tolerance and flexibility make this a more
35
appealing option for most systems. With this model, the maximum capacity p∗ is 28
nodes.
1 2 3 4 5 61
2
3
4
5
6
7
8
9
10
σ = 0.0133, κ = 0.001234
Number of Nodes
Extr
acti
onT
hro
ugh
put
(min
ofau
dio
/se
c)
MeasuredUSL model
Linear
Figure 5.9: Feature extraction cluster/node scalability (HDFS distributed audio files)
36
Chapter 6
Conclusions
Many of the current frameworks for audio feature extraction have focused on algo-
rithms and optimization techniques for serial execution. Dealing with larger datasets
requires a more scalable model for extraction. In this paper, the Safe system for
high-performance concurrent audio feature extraction was outlined. In comparison
benchmarks with current leading MIR extraction libraries, Safe outperformed the
existing solutions for all datasets with more than 3 minutes of audio.
Scalability analysis provided models for the system’s multi-processor and multi-
node cluster scalability. The actor model of concurrency used in Safe provides a simple
theoretical model for predicting the vertical (processor) scalability with any given
feature set. Further scalability testing demonstrated that the measured contention
factors lined up with the expected theoretical values.
Two different models for multi-node cluster scalability were evaluated. The first
model relied on the dataset being distributed across the cluster onto the individual
node’s local file systems. The second, more flexible and fault-tolerant model, lever-
aged a true distributed file system. Using the Universal Scalability Law to estimate
the system’s contention and coherency factors, the local file system based approach
37
was shown to have close to ideal scalability. Although slightly less performant, the
system’s scalability using the distributed file system was still very strong.
All of the source code for Safe is available online [35] and released under the
Eclipse Public License version 1.0 [43].
6.1 Future Work
In addition to adding more algorithms and features to Safe, there are certain areas
where scalability of the system can be improved. Many of the current aggregation
features in the system are un-necessarily limited to a single actor. In cases where
aggregation is performed on a per-file basis, a smarter routing component can use the
origin file information to appropriately route messages. More specifically, a consistent
hashing router can be used that takes the origin file’s path as the hash key.
All of the cluster scalability tests were performed on a homogenous cluster. Ad-
ditional research into smarter distributed routing for unbalanced or inhomogeneous
clusters would help in optimizing extraction times. Akka provides an adaptive prob-
abilistic load balancing router that can be leveraged or extended for this task.
Lastly, future enhancements should address more aspects of fault-tolerance within
the system. While minimal support for fault-tolerance is built into the system through
the libraries being used, additional logic needs to be incorporated for handling failures
during extraction. Ideally, the extraction system should be able to recover from node
failures or network partitions.
38
Appendix A
Implemented Features
A large portion of this research consists of analyzing MIR feature extraction algo-
rithms and breaking them down into smaller sequential components for dataflow
analysis. The following sub-sections describe the various feature extraction algo-
rithms implemented in the system, along with any common or shared sub-feature
computations.
A.0.1 Frame
The framing function produces an iterator of fixed-size, potentially overlapping,
frames from an audio input stream. Each frame is normalized based on the encoding
bit-depth and multiple channels are combined to produce a single frame of normalized
audio. The following algorithm shows the combination for a single sample S based
on the number of channels n and bit-depth d:
S =
∑ni=1 si
(2d − 1)n(A.1)
39
A.0.2 Window
Safe provides a common set of windowing functions, including Hann (A.2), Hamming
(A.3), Blackman (A.4), Blackman-Harris (A.5), and Bartlett (A.6) windows.
w(n) = 0.5(
1− cos(
2πn
L− 1
)), 0 ≤ n < L (A.2)
w(n) = 0.54− 0.46 cos(
2πn
L− 1
), 0 ≤ n < L (A.3)
w(n) = 0.42− 0.5 cos(
2πn
L− 1
)+ 0.08 cos
(4π
n
L− 1
), 0 ≤ n < L (A.4)
w(n) = a0 − a1 cos( 2πn
L− 1
)+ a2 cos
( 4πn
L− 1
)− a3 cos
( 6πn
L− 1
), 0 ≤ n < L
a0 = 0.35875, a1 = 0.48829, a2 = 0.14128, a3 = 0.01168 (A.5)
w(n) =
2nL−1 0 ≤ n ≤ L−1
2
2− 2nL−1
L−12≤ n < L
(A.6)
A.0.3 Fast Fourier Transform
The FFT algorithm used in Safe is a JVM-optimized version of the standard Cooley-
Turkey algorithm. The standard Cooley-Turkey algorithm is described as:
Xk =
N/2−1∑m=0
x2me−2πiN/2
mk + e−2πiN
k
N/2−1∑m=0
x2m+1e− 2πiN/2
mk (A.7)
40
Further information on the implementation and optimization techniques used can
be found online [36].
A.0.4 Power Spectrum
Safe provides various spectrum functions, including Power P (k), Magnitude A(k),
and Phase φ(k). Each function is based on the spectral coefficients ak and bk computed
by the FFT.
P (k) = a2k + b2k (A.8)
A(k) =√a2k + b2k (A.9)
φ(k) = tan−1bkak
(A.10)
A.0.5 Constant-Q Transform
The Constant-Q transform (CQT) is similar to the Fourier transform, but uses log-
arithmic spacing instead of linear spacing. This is beneficial in audio analysis since
notes in western music are spaced logarithmically. The basic CQT algorithm is given
by:
Xk =1
Nk
Nk−1∑n=0
wk,nxne−j2πQnNk (A.11)
The CQT algorithm used in Safe is based on the Brown and Puckette [6] algorithm,
which leverages the standard FFT algorithm for faster computations. Figure A.1
shows the dataflow for the CQT algorithm.
41
Audio in Frame Window FFT CQT
Figure A.1: Sequential dataflow of the Constant-Q transform
Name Description Default Value
sampleRate Target (expected) sample rate of audio inputs 44100stepSize The step size (number of samples) for the fram-
ing function512
windowType Windowing function (bartlett, blackman, black-manHarris, hamming, hann)
hann
binsPerOctave The number of CQT bins/octave (default 24 =quarter-tone spacing)
24
maxFreq Maximum frequency (hz) to look for (default isG10, MIDI note 127)
12543.854
minFreq Minimum frequency (hz) to look for (default isC1, MIDI note 12)
16.351599
threshold Minimum threshold 0.0054
Table A.1: CQT Parameters
A.0.6 Mel-Frequency Cepstral Coefficients
The mel-frequency cepstrum provides a description of an audio signals power spectrum
based on the mel scale, a perceptually driven frequency scale. In this scale, the mel
representation m of a frequency f (hz) is given by:
m = 1127 loge
(1 +
f
700
)(A.12)
The MFCC computation is based on the Davis and Mermelstein [12] algorithm,
which is described as:
Yk =
N/2−1∑f=0
A(k)wk(f) (A.13)
MFCCn =K−1∑k=0
log Yk cos
((2k + 1)nπ
2K
)
42
Where Yk represents the application of K overlapping triangular (mel-scale
spaced) filters on the magnitude spectrum of the signal A(k) and n number of MFCC
coefficients are calculated through a discrete-cosine transform on the log-energy
outputs of Yk. Figure A.2 shows the dataflow for the MFCC algorithm.
Audio in Frame Window FFT Mag MFCC
Figure A.2: Sequential dataflow for Mel-Frequency Cepstral Coefficients
Name Description Default Value
sampleRate Target (expected) sample rate of audio inputs 44100frameSize Frame size (number of samples) for the framing
function1024
stepSize Step size (number of samples) for the framingfunction
512
windowType Windowing function (bartlett, blackman, black-manHarris, hamming, hann)
hann
numCoeffs Number of cepstral coefficients to extract 13melFilters Number of mel filter banks to use 40minFreq Minimum frequency (hz) for the filter bank 130.0maxFreq Maximum frequency (hz) for the filter bank 6854.0
Table A.2: MFCC Parameters
A.0.7 Spectral Shape
Spectral shape is an aggregate of four separate spectral features: Centroid, Spread,
Skewness, and Kurtosis. The calculations for these features are described in [21], and
are defined by the following equations:
Scentroid = µ1 (A.14)
Sspread =√µ2 − µ2
1 (A.15)
43
Sskewness =2µ3
1 − 3µ1µ2 + µ3
S3spread
(A.16)
Skurtosis =−3µ4
1 + 6µ1µ2 − 4µ1µ3 + µ4
S4spread
− 3 (A.17)
Where µi =∑N−1k=0 kiA(k)∑N−1k=0 A(k)
and A(k) represents the magnitude spectrum of the signal.
Figure A.3 shows the dataflow sequence for the spectral shape algorithm.
Audio in Frame Window FFT Mag Shape
Figure A.3: Sequential dataflow for Spectral Shape
Name Description Default Value
sampleRate Target (expected) sample rate of audio inputs 44100frameSize Frame size (number of samples) for the framing
function1024
stepSize Step size (number of samples) for the framingfunction
512
windowType Windowing function (bartlett, blackman, black-manHarris, hamming, hann)
hann
Table A.3: Spectral Shape Parameters
A.0.8 Spectral Flux
The spectral flux algorithm is based on the algorithm for measuring change in the
power spectrum described by Dixon in [16]. The algorithm computes the flux between
consecutive frames (n, n − 1) as the summation of the positive differences between
the magnitude at each frequency bin, as shown in the following equation.
SF (n) =N−1∑k=0
H(|X(n, k)| − |X(n− 1, k)|
)(A.18)
44
In this equation, H represents a half-wave rectifier function where H(x) = x+|x|2
.
Figure A.4 below shows the dataflow for the spectral flux extraction algorithm.
Audio in Frame Window FFT Mag Flux
Figure A.4: Sequential dataflow for Spectral Flux
Name Description Default Value
sampleRate Target (expected) sample rate of audio inputs 44100frameSize Frame size (number of samples) for the framing
function1024
stepSize Step size (number of samples) for the framingfunction
512
windowType Windowing function (bartlett, blackman, black-manHarris, hamming, hann)
hann
diffLength Compare frames spaced n length apart, 1 = con-secutive frames
1
Table A.4: Spectral Flux Parameters
A.0.9 Onset Detection
Safe implements the spectral onset detection described in [4]. The algorithm extends
the standard spectral flux calculation by applying a logarithmically spaced filter-bank
(similar to the Constant-Q transform spacing). The onset detection function is given
by:
O(n) =
k=N/2∑k=1
H(|X log
filt(n, b)| − |Xlogfilt(n− 1, b)|
)(A.19)
Peaks in the onset detection function are selected if all of the following conditions
are satisfied:
1. ODF (n) = max(ODF (n− w1 : n+ w2)
45
2. ODF (n) ≥ mean(ODF (n− w3 : n+ w4)) + δ
3. n− nlast onset > w5
Where δ is a fixed threshold value and w1, w2, . . . , w5 are various tunable param-
eters. Figure A.5 shows the dataflow for the onset detection algorithm.
Audio in Frame Window FFT Mag
Filter Fiux Onsets
Figure A.5: Sequential dataflow for Spectral Onsets
Name Description Default Value
sampleRate Target (expected) sample rate of audio inputs 44100frameSize Frame size (number of samples) for the framing
function1024
stepSize Step size (number of samples) for the framingfunction
512
windowType Windowing function (bartlett, blackman, black-manHarris, hamming, hann)
hann
ratio Minimum activation ratio for windowing func-tion
0.22
threshold Minimum threshold (δ) for peak-picking 2.5
Table A.5: Spectral Onset Parameters
46
Tab
leB
.1:
Fu
llE
valu
ati
on
Data
set
Song
Artist
Album
Genre
MD5
Han
gin
gT
ree
Cou
nti
ng
Cro
ws
Satu
rday
Nig
hts
&..
.A
lter
nati
ve7882ab
09d
5a2330e8
6d
184410d
985d
b1
Dra
wM
eD
own
Pat
Mar
tin
oM
issi
on
Acc
om
pli
shed
Jazz
c7a198ff
c99ea
05166fe
6d
d11fe
db
461
Su
mm
erti
me
Rol
ls(
...
Jan
e’s
Ad
dic
tion
Noth
ing’s
Sh
ock
ing
Alt
ern
ati
ve35d
132b
0d
7fb
f24946fc
1d
f98f9
f437d
(For
ewor
d)
Pai
nO
fS
alva
tion
Entr
op
iaP
rog.
Met
al
a2eb
384
6951687377255b
6f1
27f9
d1b
b
2000
Lig
ht
Yea
rsA
way
Gre
enD
ayK
erp
lun
k!
Pu
nk
74523401139759d
223ad
3007f3
545b
59
3W
eeks
InL
aym
enT
erm
s3
Wee
ks
InIn
die
921855a0f3
f5d
36afd
92b
dc3
7741b
943
3rd
Pla
net
Mod
est
Mou
seT
he
Moon
&A
nta
rc..
.In
die
560c9
73b
19c1
2e2
67f5
3d
4b
544c1
73e4
Hol
iday
sJoh
nS
cofi
eld
Ou
tL
ike
AL
ight
Jazz
1819b
9f7c3
8695823fe
dee
c4151ed
3b
a
AF
eed
bag
ofT
ruck
s...
Lag
wag
onL
et’s
talk
ab
ou
t..
.P
un
kd
53f2
e3c8
fa924b
f840cd
40601a1a3a8
Ace
sH
igh
Iron
Mai
den
Pow
ersl
ave
Met
al
05018d
11e5
0f6
0f8
8861cff
9b
b5c2
cc9
Ad
dV
ice
Lov
eM
eD
estr
oyer
Bla
ckH
eart
Aff
air
Pu
nk
6d
b19b
64b
d9d
e5f9
1c4
1ec
c2a70b
28b
1
Ad
vent
Op
eth
Morn
ingri
seP
rog.
Met
al
c945b
0880f4
aef
e97a2d
7b
49d
34ca
6ac
Aft
erY
ouM
yF
rien
dL
agw
agon
Let
’sT
alk
Ab
ou
t..
.P
un
k5ca
a8e2
d3ef
29814fb
e0e7
cb7c7
37396
All
At
Sea
(Liv
eat
...
Jam
ieC
ull
um
Catc
hin
gT
ale
sJazz
75e3
ba8
b0b
c892d
6b
443f1
63c5
7b
6a33
All
For
You
Ou
rL
ady
Pea
ceG
ravit
yA
lter
nati
vef7
9d
e432e5
01c4
cf5b
a7595fd
42d
2733
An
gels
Hol
oca
ust
Iced
Ear
thN
ight
Of
Th
eS
tor.
..M
etal
2ec
8aca
5025f3
5cb
dd
ae8
d2e8
77d
0a8f
An
gels
Los
ing
Sle
epO
ur
Lad
yP
eace
Hea
lthy
InP
ara
no..
.A
lter
nati
ve35e3
20a0f7
2b
c40671f5
0e1
fb2d
d2254
An
imae
Par
tus
(’I
Am
’)P
ain
Of
Sal
vati
on
Be
Pro
g.
Met
al
ab
ff5389
ac9
6129ef
49cf
7c9
4d
0a34cb
Bab
elM
um
ford
&S
ons
Bab
elF
olk
Rock
3e2
bb
42925550b
0927cb
8231f0
dca
352
48
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Bag
ItU
pO
asis
Dig
Ou
tY
ou
rS
oul
Alt
ern
ati
vecb
ee488
a1963b
ac4
8ef
d3248b
85ae3
d4
Beg
inn
ings
Pai
nO
fS
alva
tion
Rem
edy
Lan
eP
rog.
Met
al
66a883b
72d
781714139fb
00b
15cc
717c
Bid
in’
My
Tim
eJoe
Pas
sIr
a,
Geo
rge
An
dJoe
Jazz
b423d
a7c4
c76fe
9834830b
7ab
6d
7689b
Bla
ckD
ogL
edZ
epp
elin
Led
Zep
pel
inIV
Rock
cfe7
2b
189d
267cb
3d
f70f4
e246f9
2ec
6
Bli
zzar
dO
f’7
7N
ada
Su
rfL
etG
oIn
die
d576420
f021075a870c6
95fc
1fc
32d
db
Bri
ckw
ork,
Par
t1
-...
Pai
nO
fS
alva
tion
12-5
Pro
g.
Met
al
2ed
9242
23217cc
63794e0
44b
bcc
f47a5
Bu
llio
nM
ille
nco
lin
Lif
eon
aP
late
Pu
nk
c6f5
6d
5fd
81d
255c7
1e4
cfd
317c7
6a3d
Bu
rnL
agw
agon
Bla
zeP
un
k3b
4e4
adcf
e0a755d
6d
32b
cc00044ac1
7
Bu
rnou
tG
reen
Day
Dookie
Pu
nk
3482230b
f80309c2
60b
fd2285708222e
Cau
ght
Som
ewh
ere
in..
.Ir
onM
aid
enS
om
ewh
ere
inT
ime
Met
al
6139ee
48b
d373a4161a614c7
7b
2b
402e
Ch
oked
An
dC
har
med
Lov
eM
eD
estr
oyer
Th
eT
hin
gs
Aro
un
d..
.P
un
k81d
1692
5f6
6c1
7d
962f7
3ef
6b
bef
dd
94
Cit
yB
oyB
lues
Mtl
eyC
reT
hea
tre
Of
Pain
Met
al
2a6e6
5c7
2f1
95ce
e17f5
dcc
7a51e8
6c8
Con
cret
eB
edN
ada
Su
rfT
he
Wei
ght
isa
Gif
tIn
die
0992d
43f5
050d
929e3
b4b
697795ab
223
Cow
boy
sF
rom
Hel
lP
ante
raC
owb
oys
Fro
mH
ell
Met
al
82f5
ebcd
e5ee
9b
77d
f953cc
73828cf
e9
Cra
zyT
rain
Ozz
yO
sbou
rne
Th
eE
ssen
tial
Ozz
...
Rock
483d
7c8
b54313a2332d
933ab
24a0694e
Cro
pC
ircl
eM
onst
erM
agn
etP
ower
Tri
pS
ton
erM
etal
b049538
04746160cc
89f6
28d
095e5
326
Cry
Tou
ghP
oiso
nL
ook
Wh
at
Th
eC
at.
..M
etal
2f8
74d
b33cf
cbc2
0469f7
4c1
372c2
247
Cycl
ops
Rev
olu
tion
Mon
ster
Mag
net
Su
per
jud
ge
Sto
ner
Met
al
fc60f8
e77d
ebf9
c7ca
fb43e1
33b
17b
67
D’Y
ouK
now
Wh
atI
M..
.O
asis
Be
Her
eN
owA
lter
nati
ve3aaf9
ab
947c5
58154537cb
5c4
8f5
2b
7b
Dan
ny’s
Son
g(L
oggi
...
Me
Fir
stA
nd
Th
eG
i...
Hav
eA
Ball
Pu
nk
31f6
93d
ad
d240d
890b
3461e8
09d
08ce
b
49
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Dar
kS
aga
Iced
Ear
thT
he
Dark
Saga
Met
al
b5ea
1ee
fb2aee
0e4
74d
fa91b
b79acf
63
Day
Job
Gin
Blo
ssom
sC
on
gra
tula
tion
sI.
..In
die
9d
cae0
9a35fd
efb
f4ab
30d
e0a57ef
bad
Des
ert
Skie
sL
aym
enT
erm
sD
rive
To
Now
her
e-..
.In
die
181a33d
a8a378a78042e4
8fe
c65a4a57
Dis
app
ear
Maz
zyS
tar
Am
on
gM
yS
wan
Alt
ern
ati
ve50901c3
4a28584fd
1c2
7114345cf
69c8
Doi
ng
ItF
orT
he
Kid
sN
obod
ys
Gen
erati
on
XX
XP
un
k4b
f4a35
c41a8ed
21b
72c1
d65e0
7d
e445
Dop
esT
oIn
fin
ity
Mon
ster
Mag
net
Dop
esT
oIn
fin
ity
Sto
ner
Met
al
2590f9
b0c6
ea1a7fd
cb88f9
2338d
55e8
Fad
eIn
toY
ouM
azzy
Sta
rS
oT
on
ight
Th
at
I...
Alt
ern
ati
veafc
6c7
db
005538a9a305a2b
2a34e2
b21
feve
rM
ich
ael
Bu
ble
Wit
hL
ove
Jazz
e15c2
f7610f1
575336e6
27c3
94f0
db
c4
Fol
low
Th
rou
gh(S
tr..
.G
avin
DeG
raw
Ch
ari
ot
-S
trip
ped
Pop
Rock
0c8
acb
8cb
883b
730fd
a713580a670a94
Fol
low
Th
rou
ghG
avin
DeG
raw
Ch
ari
ot
Pop
Rock
ac7
45e9
d798d
527aa9f7
8882b
886cc
e2
For
ce-F
edG
od
For
bid
Gon
eF
ore
ver
Met
al
27334a7ec
b14f3
5c1
27ec
cfcc
9c6
078e
For
get
Her
Jeff
Bu
ckle
yG
race
Dis
c2
Ind
iec4
e429fa
ebeb
ae6
feb
166190d
c0a6b
90
Fro
mT
he
Roof
top
Dow
nN
ada
Su
rfT
he
Wei
ght
IsA
G..
.In
die
a56ed
273d
62e7
dc5
377a30cb
97d
32f7
0
F**
*in
’In
Th
eB
ush
esO
asis
Sta
nd
ing
On
Th
eS
...
Alt
ern
ati
vec3
460c0
0ed
084b
d6686b
99b
e81111584
Fu
ture
sJim
my
Eat
Worl
dF
utu
res
Ind
ie0226a02155098c1
8d
2e7
4fd
327cc
4cb
3
Gas
olin
eK
isse
sF
or..
.P
ark
ItW
on
’tS
now
Wh
e...
Pu
nk
c419cc
df4
75fd
79a12ce
3ac4
10f2
0a49
Giv
em
eth
eN
ight
Geo
rge
Ben
son
Ver
yb
est
of
-G
r...
Jazz
48b
9935
c31d
4c1
4398517a4064a5e7
8a
Go
Pea
rlJam
Vs.
Alt
ern
ati
ve6089d
775728ed
129690b
56109a4b
bee
1
God
Kn
ows
IL
ove
Yo.
..L
oren
eD
rive
Rom
anti
cW
ealt
hP
un
k17b
7502
9b
08b
0f6
378fc
22c9
ad
07b
27b
Gon
na
See
My
Fri
end
Pea
rlJam
Back
space
rA
lter
nati
ve109ce
76eb
f124f8
d0ee
e277afd
dd
ab
e2
50
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Good
Tim
esB
adT
imes
Led
Zep
pel
inL
edZ
epp
elin
Rock
03d
1519
d4b
58180e3
00fd
247a49c2
46d
Intr
oG
oril
laz
Dem
on
Day
sH
ip-H
op
cf197cd
ef57e0
b8ce
0ca
7aef
5ab
7e3
a9
Hea
rtis
Har
dto
Fin
dJim
my
Eat
Worl
dIn
vente
dIn
die
a23d
7f7
65afa
6ad
1b
dd
7e0
91b
dd
9b
345
Hea
rtb
reak
ing
Mu
sic
Lag
wag
onR
esolv
eP
un
k4f7
e11833e3
b39e3
de0
db
0cb
ce78c4
cf
Hel
loO
asis
(Wh
at’
sT
he
Sto
ry..
.A
lter
nati
ve7391d
6e0413f9
a5c4
149b
41ec
dd
0754b
Hol
yF
***
IA
mT
he
Ava
lan
che
Ava
lan
che
Un
ited
Pu
nk
7986ee
f40b
4c4
20e9
5a53e1
d277f9
268
Hol
yW
ars.
..T
he
Pu
...
Meg
adet
hR
ust
InP
eace
Met
al
7515b
288ed
22a3302458e3
00b
ad
c49ca
How
We
Get
Alo
ng
Ju
rass
ic5
Qu
ali
tyC
ontr
ol
Hip
-Hop
2a9b
bb
93f3
edad
1cd
c288e6
5781c9
075
ID
on’t
Kn
owO
zzy
Osb
ourn
eB
lizz
ard
of
Ozz
Rock
631f5
1c7
5e8
78043ea
8ce
42786d
da81e
Imm
igra
nt
Son
gL
edZ
epp
elin
Led
Zep
pel
inII
IR
ock
0b
ce4fa
bc4
7aa3b
e3a40d
c2aef
61f8
12
InT
he
Eve
nin
gL
edZ
epp
elin
InT
hro
ugh
Th
eO
u..
.R
ock
d62e0
c9c0
6c3
5849807ea
480926b
126c
InT
he
Mea
nti
me
Hel
met
Mea
nti
me
Alt
ern
ati
ve7fd
e7ea
084ec
9b
21ce
89cc
6769cb
901f
Ind
epen
den
ceD
ayN
oM
otiv
Day
light
Bre
akin
gP
un
ke0
8d
820
6c3
c54e7
3f6
dad
a8031d
f1a9d
Into
You
rE
yes
Lu
cero
Th
eA
ttic
Tap
esA
lt-C
ou
ntr
y1769f3
f2ca
9cd
c155cb
ec363d
1e5
4b
64
Intr
od
uct
ion
Pan
ic!
At
Th
eD
isco
AF
ever
You
Can
’t...
Pu
nk
f9680585c8
f690468f3
07e9
dd
977452d
Inva
der
sIr
onM
aid
enT
he
Nu
mb
erO
fT
he.
..M
etal
5395cc
d950fb
6598b
76684b
0cd
7a91f1
Wh
atA
Diff
eren
ceA
...
Jam
ieC
ull
um
Tw
enty
som
eth
ing
Jazz
b73afa
bf4
38c6
6fc
957905471c2
0f4
46
Jaw
sof
Lif
eJoh
nP
etru
cci
Su
spen
ded
An
imati
on
Pro
g.
Met
al
a1d
a6fc
b4622c6
7a22f3
77ef
2ee
77f0
4
Ju
stO
ne
Of
Th
ose
T..
.Jam
ieC
ull
um
Th
eP
urs
uit
Jazz
81b
e1cf
ab
53c8
6b
d37025f6
7950967af
Ju
stifi
edB
lack
Eye
No
Use
For
AN
am
eL
ech
eC
on
Carn
eP
un
k268f9
0c6
9ec
7d
d57ce
71b
5665b
3ea
ee2
51
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Kid
sD
on’t
Lik
eT
o..
.L
agw
agon
Hoss
Pu
nk
c4d
9d
0425d
9331b
2b
c471e7
fd32715d
4
Las
tof
the
Anti
-Fa.
..K
ane
Hod
der
Th
eP
leasu
reto
R..
.P
un
kf7
fc3f6
4ef
2b
7a60e4
ba61293ec
5e9
81
Lif
eon
aC
hai
nP
ete
Yor
nm
usi
cfort
hem
orn
in..
.A
lter
nati
ve989580a25ef
66c9
b52ea
6cc
d57d
25793
Lif
eW
aste
dP
earl
Jam
Pea
rlJam
Alt
ern
ati
veb
29e9
79b
a2b
aa9d
95f3
f0d
343c7
0f3
9d
Lin
oleu
mN
OF
XP
un
kIn
Dru
bli
cP
un
k5b
d99e5
956d
ca06c7
14b
3a040d
678d
80
Lit
tle
Dip
per
Hu
mY
ou
’dP
refe
ran
A...
Alt
ern
ati
ve16538e3
b8d
2b
49645e8
7e4
80b
62b
35f5
Liv
ing
Aft
erM
idn
ight
Ju
das
Pri
est
Bre
akin
gth
eL
awM
etal
14ce
49d
bd
5b
a191f7
112e7
cc67a9f4
54
Los
tH
oriz
ons
Gin
Blo
ssom
sN
ewM
iser
ab
leE
xp
...
Ind
ie1c2
b4ee
7c5
a65334e1
7d
0265b
251a2ce
Mam
a,I’
mC
omin
gH
ome
Ozz
yO
sbou
rne
Th
eE
ssen
tial
Ozz
...
Rock
d604c7
97ae6
9ab
1fc
cf6128d
a2e0
e425
Mat
tS
kib
a-
Good
F..
.M
att
Skib
aS
pli
tIn
die
da9145f
936209f0
b2ed
6c1
0649e8
845e
Mel
tM
onst
erM
agn
etG
od
Say
sN
oS
ton
erM
etal
8c8
e99fc
ec6949fe
71a5ab
2d
fd23ff
1a
Mo
joP
inJeff
Bu
ckle
yG
race
Dis
c1
Ind
ie694848469f3
fcb
83a0fe
ed9d
51f4
b6cf
Mou
thF
orW
arP
ante
raV
ulg
ar
Dis
pla
yof.
..M
etal
f4c2
dd
7063cd
e0e7
ac6
617a2cb
dac7
19
Ele
ctro
cuti
onN
ada
Su
rfIf
IH
ad
aH
i-F
iIn
die
a8af0
64b
061132493103001934b
dd
cdf
New
Dir
ecti
onG
oril
laB
iscu
its
Sta
rtT
od
ayP
un
kaf1
4b
030b
1a5e9
325542e8
0a94d
a7a27
Nig
htm
ares
by
the
Sea
Jeff
Bu
ckle
yS
ketc
hes
for
My
S..
.In
die
db
3f2
e02f1
6d
2187549d
dd
b6d
9ac0
675
Div
eN
irva
na
Ince
stic
ide
Alt
ern
ati
ved
57e3
f21444445f8
a6d
537295d
13679f
Ser
veT
he
Ser
vants
Nir
van
aIn
Ute
roA
lter
nati
ve77ea
1b
418d
3594423988ad
9174762d
37
Sm
ells
Lik
eT
een
Sp
...
Nir
van
aN
ever
min
dA
lter
nati
ve2d
5c0
46e6
6ce
e3b
f4b
aef
48ad
9d
2c3
d7
No
Cig
arM
ille
nco
lin
Pen
nyb
rid
ge
Pio
nee
rsP
un
k8472f6
ad
2e9
5cd
19e7
bc7
2f9
4b
4760b
7
52
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Acq
uie
sce
Oas
isT
he
Mast
erp
lan
Alt
ern
ati
vef5
2ff
6c7
f62996ff
0c0
784fa
b881a678
On
ceP
earl
Jam
Ten
(Del
uxe
Ed
itio
n)
Alt
ern
ati
ve0638cf
b17485b
b98874a76b
bb
ec1b
b40
On
eM
anA
rmy
Ou
rL
ady
Pea
ceH
ap
pin
ess.
..Is
N..
.A
lter
nati
ve59f9
bf2
708af9
d6d
7656ef
be9
25245b
9
On
eQ
uie
tN
ight
Pat
Met
hen
yO
ne
Qu
iet
Nig
ht
Jazz
049d
0a2
05f5
a6f9
a7f9
947107915cd
33
Orc
hes
tral
Intr
o(f
...
Gor
illa
zP
last
icB
each
Hip
-Hop
9fd
734c
1a4ef
52016c3
70560ee
cd03d
f
Ove
rT
he
Rai
nb
ow(W
...
Me
Fir
stA
nd
Th
eG
i...
Are
AD
rag
Pu
nk
93c2
c56e0
e0663ad
caab
8a61cb
12b
077
Pau
seK
idD
yn
amit
eK
idD
yn
am
ite
Pu
nk
16541e8
7688202b
dcc
255ab
38a10b
a92
Pea
cefu
lD
ayP
ennyw
ise
Ab
ou
tT
ime
Pu
nk
4d
6aee
60436141b
a1ea
d7d
28f1
f9a8c7
Bre
aker
fall
Pea
rlJam
Bin
au
ral
Alt
ern
ati
ve5d
6732b
76d
5d
81e1
c940355b
e1c9
d213
Can
’tK
eep
Pea
rlJam
Rio
tA
ctA
lter
nati
ved
b1cc
e4b
f7e3
2571b
ea3318cf
9098ed
c
Las
tE
xit
Pea
rlJam
Vit
alo
gy
Alt
ern
ati
ve3a6ea
d2d
99d
a072b
de6
96c5
2f3
f5f5
65
Pil
lS
hov
elM
onst
erM
agn
etS
pin
eof
God
Sto
ner
Met
al
173c4
2e7
3a3cf
1063618d
e7541b
db
606
Pit
sA
nd
Poi
son
edA
...
Kid
Dyn
amit
eS
hort
er,F
ast
er,L
o..
.P
un
ke9
52c3
f26ab
0cf
12d
1a6d
c5a4212b
837
Pu
reH
elm
etA
fter
tast
eA
lter
nati
ve9d
90ab
d1c6
4347c5
988a75458ad
fba77
R.K
.In
tro
Ou
rL
ady
Pea
ceS
pir
itu
al
Mach
ines
Alt
ern
ati
vead
3d
70a679991e3
8ad
93e0
560ec
b6486
Re-
Has
hG
oril
laz
Gori
llaz
Hip
-Hop
3d
79f7
f00e1
04798cb
f8685c5
af3
a827
Rem
ind
erG
rey
Are
aF
anb
elt
Alg
ebra
Pu
nk
d1a4aee
1b
6994a13049881716c1
ceaab
Rep
etit
ion
Hel
met
Str
ap
ItO
nA
lter
nati
ved
2d
f5b
029c5
52c7
27b
63a1081c6
76893
Rig
ht
Now
!G
rey
Are
aG
reya
rea
Pu
nk
af3
0aea
5b
f868a9ef
00f9
c8d
89a16d
f6
Rip
ItO
ut
Kis
sA
ceF
reh
ley
Rock
c2a011ee
7360b
01ef
112f9
414e4
d71cd
53
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Rock
’N’
Rol
lS
tar
Oas
isD
efin
itel
yM
ayb
eA
lter
nati
veef
874b
edfb
5f2
b73ef
06005681f3
1b
d6
Ru
by
Kai
ser
Ch
iefs
You
rsT
ruly
,A
ngr.
..In
die
a996a7415b
6d
e75a3e2
14a9640a7776c
Ru
mb
leY
oun
gM
anR
u..
.P
inh
ead
Cir
cus
Th
eB
lack
Pow
ero..
.P
un
kc5
0cd
aacb
b5cf
9a5a11e7
eab
915b
ac2
8
Sal
tS
wea
tS
uga
rJim
my
Eat
Worl
dB
leed
Am
eric
an
Ind
ie44e3
dcc
41606c0
f24b
a8d
2b
be8
e7a0b
8
Sat
elli
teS
kin
Mod
est
Mou
seN
oO
ne’
sF
irst
,A
...
Ind
ieaf7
0514b
f6616337cf
cfe2
d06cf
dd
6d
4
Sca
rsic
kP
ain
Of
Sal
vati
on
Sca
rsic
kP
rog.
Met
al
8b
df1
4b
7c1
aa06c5
ccd
a29d
bb
e36b
e2a
Scr
eam
ing
For
Ven
ge..
.Ju
das
Pri
est
Met
al
Work
s’7
3’.
..M
etal
606191afc
0ad
9aa19e2
1d
784236ca
3a9
See
Th
eW
ind
Pol
arB
ear
Clu
bC
hasi
ng
Ham
bu
rgP
un
k4b
86cb
11b
1a072c8
0d
3a4e9
fa4c8
85e2
Sh
adow
sO
fD
efea
tG
ood
Rid
dan
ceO
per
ati
on
Ph
oen
ixP
un
k8836205c1
ff8b
b8b
6afd
feeb
1f3
e1aa5
Sh
amb
les
Mak
eD
oA
nd
Men
dB
od
ies
Of
Wate
rP
un
k95e0
94e0
f6ad
eab
a164c9
7c5
ec9d
2e0
5
Sig
hN
oM
ore
Mu
mfo
rd&
Sons
Sig
hN
oM
ore
Folk
Rock
c05ce
9f5
29280072b
89f9
4a9c8
8d
98b
0
Skin
O’
My
Tee
thM
egad
eth
Cou
ntd
own
To
Exti
...
Met
al
c7b
ea38
6a7c9
c40412a7a376202b
9614
Slu
tM
ach
ine
Mon
ster
Mag
net
Mon
oli
thic
Baby!
Sto
ner
Met
al
6757d
7ca7c4
3ae8
6c6
427153527b
db
fa
Sm
art
Hel
met
Siz
eM
att
ers
Alt
ern
ati
ve95866e0
d750e5
71710b
086f5
29d
ffcb
8
Sm
oke
Lu
cero
1372
Over
ton
Park
Alt
-Cou
ntr
y918f5
2f8
deb
2963111e8
d79174f8
bf8
3
Sn
ake
Dan
ceM
onst
erM
agn
etM
on
ster
Magn
etS
ton
erM
etal
f939675835166102fa
fb507ff
da2d
f02
So
Lon
gH
elm
etS
eein
gE
ye
Dog
Alt
ern
ati
vea84d
6c9
540c0
26b
5d
718fc
f365f0
ad
4f
So
Wh
atG
eorg
eB
enso
nB
eyon
dT
he
Blu
eH
...
Jazz
18a438b
d509ce
76ae0
4d
4d
952822584e
Som
etim
esP
earl
Jam
No
Cod
eA
lter
nati
ved
cfec
22ef
117528e6
81271722380810c
Sp
eak
To
Me
/B
reat
he
Pin
kF
loyd
Dark
Sid
eof
the
...
Rock
37b
d9e9
2c0
006f3
db
005d
dc1
d8f9
e94a
54
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Sp
irit
ofth
eL
and
Pai
nof
Sal
vati
on
On
eH
ou
rby
the
C..
.P
rog.
Met
al
634ae3
fd140915d
231b
f688e7
d146ec
6
Su
per
man
’sD
ead
Ou
rL
ady
Pea
ceC
lum
syA
lter
nati
vea21aa4c5
a2b
daf4
d5443e6
d20c7
660c8
Sw
allo
win
gE
very
thin
gH
elm
etM
on
och
rom
eA
lter
nati
ve35f3
a1564481b
68554a88c8
8b
c7ff
dc9
Sw
eet
Lit
tle
Th
ing
Lu
cero
Ten
nes
see
Alt
-Cou
ntr
ycc
0d
251
979b
f311cc
2b
f9c9
082c9
9243
Sw
eet
Hom
eA
lab
ama
Lynyrd
Skynyrd
Gold
(Rem
ast
ered
)R
ock
70c2
eb2b
4c9
4a33575c6
e1e9
10635f9
6
T.n
.T.
(Ter
ror
’nT
...
Mtl
eyC
reD
r.F
eelg
ood
Met
al
745773416fb
da3fb
04f7
f11d
480b313e
Tab
...
Mon
ster
Mag
net
Tab
...2
5S
ton
erM
etal
cf036f5
acc
3012f1
06ce
cc14354d
5721
Tab
leF
orG
lass
esJim
my
Eat
Worl
dC
lari
tyIn
die
6c3
84a9b
9063406cd
8f4
710f1
8f2
03a3
Th
anks
for
the
Kil
l...
Min
us
the
Bea
rH
igh
lyR
efin
edP
i...
Ind
ie1b
11753
1a7f3
15b
7046ef
be0
6d
731f2
6
Th
atM
uch
Fu
rth
erW
...
Lu
cero
Th
at
Mu
chF
urt
her
...
Alt
-Cou
ntr
yd
0ea
886
8fc
d2e9
334f9
ffc5
4a9ed
385c
Th
atM
uch
Fu
rth
erW
est
Lu
cero
Th
at
Mu
chF
urt
her
...
Alt
-Cou
ntr
yd
8d
2b
53e9
b95b
84fa
75d
44fd
d4891ab
9
Th
eB
ird
man
Ou
rL
ady
Pea
ceN
avee
dA
lter
nati
ve7312438a18174ed
944c3
c37c1
02af5
ea
Th
eB
oat
Dre
ams
fro.
..Jaw
bre
aker
24
Hou
rR
even
ge
T..
.P
un
kd
c9b
5e1
fb5fd
46636aa4f5
57b
507d
764
Th
eF
row
nin
gof
aL
...
Hey
Mer
ced
esE
very
nig
ht
Fir
eW
...
Ind
ie17a7d
afb
fa07ad
d20765d
ed7f3
aaad
aa
Th
eG
ame
Nee
ded
Me
Min
us
Th
eB
ear
Men
os
El
Oso
Ind
ie668aa697d
883d
30b
6b
7fd
dcb
5b
486a72
Th
eG
hos
tY
ouA
reP
ark
No
Sig
nal
Pu
nk
81413b
f9577090d
1a5c3
bd
7c3
bad
e10e
Th
eH
elli
onJu
das
Pri
est
Met
al
Work
s’7
3-’
...
Met
al
73f0
9a25376d
e132cf
0ab
387549f3
e38
Th
eH
ind
uT
imes
Oas
isH
eath
enC
hem
istr
yA
lter
nati
vea08333248b
ebb
54ad
28d
7043ff
5fd
bd
2
Th
eS
ky
Isa
Lan
dfi
llJeff
Bu
ckle
yS
ketc
hes
for
My
S..
.In
die
cca31e7
58e4
cffe3
c9d
aa1d
1b
dc1
2e3
2
Th
eS
ong
Rem
ain
sT
h..
.L
edZ
epp
elin
Hou
ses
Of
Th
eH
oly
Rock
56b
5839
857d
02d
c8fe
22fa
9fb
4004fa
2
55
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Tip
You
rB
arte
nd
erG
lass
jaw
Wors
hip
&T
rib
ute
Pu
nk
00474572a89c7
28d
9c1
fb9344b
afc
00f
Tru
eN
atu
reJan
e’s
Ad
dic
tion
Str
ays
Alt
ern
ati
ve2650462909d
c31cb
ad
5d
9e2
a5d
a4566a
Tru
stM
egad
eth
Cry
pti
cW
riti
ngs
Met
al
67c6
a883f0
fc2c4
771c8
a2ef
c4b
25e8
f
Un
kn
owin
gly
Str
ong
Mak
eD
oA
nd
Men
dE
nd
Mea
sure
dM
ile
Pu
nk
fbad
2d
86b
b7ec
210d
520d
bf0
d82ad
506
Use
dP
ain
Of
Sal
vati
on
Th
eP
erfe
ctE
lem
e...
Pro
g.
Met
al
4608d
c0f9
ff4a0359e6
a291d
0d
5ea
e36
Wal
kD
on’t
Ru
nJoh
nny
Sm
ith
Kale
idosc
op
eJazz
ee6730a7d
6d
2e3
315589e3
a78e3
ec49c
Wav
esG
uth
rie
Gov
an
Ero
tic
Cakes
Pro
g.
Met
al
d4fc
92c
c644305484d
3116c6
fd9e9
07c
Wh
at’s
You
rN
ame
Lynyrd
Skynyrd
Str
eet
Su
rviv
ors
Rock
ecc9
9ac0
1b
7506f8
23c0
22c8
3fe
dc9
ee
Wh
ats
You
rN
ame
Lynyrd
Skynyrd
Gold
(Rem
ast
ered
)R
ock
23082a884b
0fd
85d
ea6b
e8f8
cd201789
Wh
ere
Eag
les
Dar
eIr
onM
aid
enP
iece
Of
Min
dM
etal
690461920c7
fae0
d140aff
451ce
778d
8
Wh
ole
Lot
taL
ove
Led
Zep
pel
inL
edZ
epp
elin
IIR
ock
2e5
e7fc
2c8
e3d
ead
04d
bf0
fda994566e
Wil
dS
ide
Mtl
eyC
reG
irls
,G
irls
,G
irls
Met
al
b936570
b3106a5b
6a7b
eba75711992a4
You
Kn
owY
ou’r
eR
ight
Nir
van
aN
irva
na
Alt
ern
ati
ve5746991922a151ee
29fd
b48c0
8e3
ec9a
You
Mov
eM
eH
ello
Th
eG
lam
Sin
gle
s..
.R
ock
6b
e0785
8918d
b4123d
73a0612d
caa916
Bel
lsH
eyM
erce
des
Hey
Mer
ced
esE
PIn
die
c1355228219212ae1
4e1
d6001c1
99ee
a
Dea
dA
nd
Gon
eI
Am
Th
eA
vala
nch
eI
Am
Th
eA
vala
nch
eP
un
k4b
7d
efc4
9d
437f7
b47b
08cb
9311f9
bd
8
Dra
mam
ine
Mod
est
Mou
seT
his
IsA
Lon
gD
r...
Ind
ie6b
df2
40cb
978870ce
2ce
8aec
36d
06c1
c
Au
tom
atic
Syst
emat
i...
Gar
bag
eN
ot
You
rK
ind
Of
...
Alt
ern
ati
ve4fc
346ed
c7b
2993931c3
810f2
1973ec
5
Rou
lett
eS
yst
ems
Hey
Mer
ced
esU
norc
hes
trate
dE
PIn
die
6ec
f141b
da642203cc
0169026d
f3ad
9c
Lit
tle
Sil
ver
Hea
rtL
uce
roL
uce
roA
lt-C
ou
ntr
ya277665c7
c32f9
f2459554c6
af6
1043a
56
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Mar
chin
toth
eS
eaM
od
est
Mou
seW
eW
ere
Dea
dB
efo..
.In
die
936c4
45a20b
5b
2e7
8657455ae5
77e5
fa
Hal
ahM
azzy
Sta
rS
he
Han
gs
Bri
ghtl
yA
lter
nati
vec7
2c6
6036d
6f4
03b
efa6528ce
38628d
2
Fee
lin
gG
ood
Mic
hae
lB
ub
leIt
’sT
ime
Jazz
b819d
22452066ff
255d
ad
48074110d
60
Nev
erE
ndin
gM
ath
E..
.M
od
est
Mou
seB
uil
din
gN
oth
ing
...
Ind
iea8ea
5b
cbb
7cb
8687d
57b
ae2
ad
01f6
e1d
Dir
tyP
aws
Of
Mon
ster
san
dM
enM
yH
ead
Isan
An
imal
Ind
ie83961d
d90b
5b
9fe
bff
caea
786425d
001
Old
Dev
ilM
oon
Jam
ieC
ull
um
Hea
rdIt
All
Bef
ore
Jazz
ee18120ee
e9837b
78525b
d729f2
aeb
7f
Ph
oner
To
Ari
zon
aG
oril
laz
Th
eF
all
Hip
-Hop
2382ce
1e1
b4b
434aee
7881b
dd
b322384
Qu
alit
yR
even
geA
t..
.H
eyM
erce
des
Lose
sC
ontr
ol
Ind
ie699f0
3d
b7954678a83b
ca97623f7
02ee
See
Th
ese
Bon
esN
ada
Su
rfL
uck
yIn
die
ad
b2d
841a96f3
099b
51a67d
81d
5d
e58d
Su
nsp
ots
Mod
est
Mou
seT
he
Fru
itT
hat
At.
..In
die
584c4
55138ee
c097c3
58cf
91b
cca2ef
a
Tee
thL
ike
God
’sS
h..
.M
od
est
Mou
seT
he
Lon
esom
eC
row
...
Ind
ie329a5d
6a45039092f6
62c4
ebf7
b1cc
e9
Th
eF
ire,
Th
eS
teel
...
Hot
Wat
erM
usi
c”T
he
Fir
e,T
he
St.
..P
un
kb
dac7
e9f5
fb9204c4
5264591f5
dd
6615
Ton
ight
Ain
’tG
onn
a...
Lu
cero
Dre
am
ing
InA
mer
ica
Alt
-Cou
ntr
y8d
d8e1
456d
5fc
1803a494f7
297d
5fb
b1
Tu
rnU
pT
he
Su
nO
asis
Don
’tB
elie
ve
Th
e...
Alt
ern
ati
ve54f9
77b
fe610ec
2b
1e7
519aea
7183fd
1
Wat
chIt
Bu
rnL
uce
roN
ob
od
y’s
Darl
ings
Alt
-Cou
ntr
y2aa45680e5
f138c1
e5d
acc
c9a9931ca
e
Wh
atE
lse
Wou
ldY
ou..
.L
uce
roR
ebel
s,R
ogu
es&
...
Alt
-Cou
ntr
yac1
a6a9f0
09151d
6512496336539ec
3c
Wil
lfu
lS
usp
ensi
on..
.M
od
est
Mou
seE
very
wh
ere
An
dH
i...
Ind
ied
5ab
9f0
993b
cfe9
b08f4
56d
b7d
205d3b
Wor
ms
vs.
Bir
ds
Mod
est
Mou
seS
ad
Sap
py
Su
cker
Ind
ied
4d
dcd
1622019750148067341fd
90c3
a
220
Yea
rsH
otW
ater
Mu
sic
Fu
elfo
rth
eH
ate
...
Pu
nk
a9c7
aac2
9f2
f4e4
bb
cde5
5ad
12d
906b
f
AF
ligh
tan
da
Cra
shH
otW
ater
Mu
sic
AF
light
an
da
Cra
shP
un
ka575c1
f1e0
8406178653418b
a2f8
6c9
8
57
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
AF
oxxy
Chic
kJoe
Pas
sE
xim
iou
sJazz
39a89330b
0fd
5d
5e9
0c5
37cb
6b
e9e4
6e
AM
apof
the
Wor
ldP
atM
eth
eny
AM
ap
of
the
Worl
dJazz
fb43fb
7863599b
29475971aeb
71cd
55e
AS
tree
tcar
Nam
edD
...
Ken
ny
Bu
rrel
l’R
ou
nd
Mid
nig
ht
Jazz
547ec
9d
449082749862141273aca
7ea
2
Aco
ust
icB
lues
Joe
Pas
sV
irtu
oso
No
4D
isc
2Jazz
c712512b
ec6b
53d
709b
72c2
3a38ee
0c2
Ala
chu
aH
otW
ater
Mu
sic
Nev
erE
nd
erP
un
k373cf
5d
7d
c530b
d2e4
f5e2
584c4
b0826
All
You
Did
Was
Sav
...
Ou
rL
ady
Pea
ceB
urn
Bu
rn(D
elu
xe.
..A
lter
nati
ve4188772d
888675e4
def
57313a180801c
An
gel
Wit
hT
he
Blu
esK
evin
Eu
ban
ks
Pro
mis
eO
fT
om
orr
owJazz
636305b
470e2
0b
fe1b
79fd
e0fd
7c9
c1b
Big
Cas
ino
Jim
my
Eat
Worl
dC
hase
Th
isL
ight
Ind
iec7
5d
276
8cb
2049cc
b9ec
821a78ae7
d09
Bit
ters
wee
tJim
my
Bru
no
Con
cord
Jazz
Gu
it..
.Jazz
987694d
6192f4
bec
a8ed
ea5c7
48ff
1d
a
Blo
od
An
dT
hu
nd
erM
asto
don
Lev
iath
an
Sto
ner
Met
al
ad
39177
6917d
e576c4
e43c0
f49a432a2
Cel
ebra
teN
oM
otiv
Dia
gra
mfo
rH
eali
ng
Pu
nk
150b
40c
a6c8
f4ed
9a47fc
ebf7
e592544
Ch
eek
toC
hee
kO
scar
Pet
erso
nT
he
Son
gIs
You
Jazz
2477869aa702a0fd
3ca
aa866628b
9cd
e
Ch
itli
ns
con
carn
eK
enny
Bu
rrel
lM
idn
ight
Blu
eJazz
eff46af9
4899c4
4d
650d
546a3b
3d
6e4
b
Cir
ibir
ibin
Har
ryJam
esH
arr
yJam
esJazz
f6f3
a1c4
1008b
bc9
20727741b
4f1
8403
Com
eR
ain
Or
Com
eS
...
Joh
nny
Sm
ith
Th
eS
ou
nd
Of
Th
e..
.Jazz
6453119137792693a92a49e4
069ab
a8b
Con
tras
tsJim
my
Dor
sey
Jim
my
Dors
eyJazz
456a4ec
3ef
6b
e783c3
1842e6
7c3
10f6
8
Cru
sher
Des
troy
erM
asto
don
Rem
issi
on
Sto
ner
Met
al
cbf6
ec05840b
1b
1728d
80177cd
6a9ec
d
Dow
nst
airs
Ken
ny
Bu
rrel
lG
uit
ar
Form
sJazz
03e5
9b
223c3
7fc
c6fe
41c4
3c7
73665b
8
Dr.
Jac
kle
Mil
esD
avis
Mil
esto
nes
Jazz
7e0
def
c50ed
55217a09f9
edea
0424a5b
E.V
.Jim
my
Bru
no
Lik
eT
hat
Jazz
2b
8d
0e6
fec5
0d
d29675e1
1544d
9661a4
58
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Lu
cky
Str
eet
Go
Rad
ioL
uck
yS
tree
tP
un
k45b
f0a7
ce11e4
21e9
afb
504640d
5e2
16
Wh
enD
ream
ing
Get
s..
.G
oR
adio
Do
Over
sA
nd
Sec
o..
.P
un
ka8c9
b90
cdc1
ebd
3ec
b69a7a91093b
b1a
Hol
dO
n,
Hol
dT
ight
Lay
men
Ter
ms
Sin
ceL
ast
Dec
emb
erIn
die
001a9636382a78ca
17ea
832b
f59617f1
Hor
nIn
tro
Mod
est
Mou
seG
ood
New
sF
or
Peo
...
Ind
ie328a9a53a8d
4a6fb
d75473ec
f9a26503
Mai
nli
ne
Hot
Wat
erM
usi
cE
xis
ter
Pu
nk
6e2
e8b
c3f3
500fe
0d
085fc
e9aea
8611b
IO
nly
Hav
eE
yes
fo..
.O
scar
Pet
erso
nT
he
Son
gIs
You
D..
.Jazz
65559638e1
0c8
45a9a323b
cfe5
9cb
3cf
I’m
Old
Fas
hio
ned
Joh
nny
Sm
ith
Leg
end
s:Solo
Gu
i...
Jazz
84ab
a68
f896061c4
057657036225d
7c0
Imag
ine
Joh
nL
enn
onL
enn
on
Leg
end
:T
h..
.R
ock
fceb
c7850f3
03c1
53c5
13773fe
301452
InA
Mel
low
Ton
eH
erb
Ell
isA
rriv
al
Dis
c2
Jazz
82f0
6d
ee9cf
a9923c9
3fd
979e3
872a1c
InT
he
Mood
Gle
nn
Mil
ler
Gle
nn
Mil
ler
Jazz
dff
8ee
35a5a8786053ac6
651b
8d
ce017
Iron
Cla
dL
ouH
um
Ele
ctra
2000
Alt
ern
ati
vea47e4
b427f9
116f6
cad
c262af2
5cd
e39
Isle
ofth
eC
hee
tah
Hu
mD
ownw
ard
IsH
eave
...
Alt
ern
ati
ve4ee
e70a7a3b
d0d
97fd
edc2
3b
f5cd
a640
Op
ener
Jim
my
Eat
Worl
dS
ingle
sIn
die
539c2
01cb
df9
9d
8c6
aa7fd
55f7
8af9
9d
Th
inkin
g,T
hat
’sA
llJim
my
Eat
Worl
dS
tati
cP
reva
ils
Ind
iec1
7491363a1a57fe
f97e3
35b
841e0
3c8
Look
for
the
Sil
ver.
..H
erb
Ell
isA
rriv
al
Dis
c1
Jazz
b3c7
8a9
f107a9c1
d6b
9a61d
9174d
91b
e
Dow
nto
wn
(Intr
o)L
uce
roW
om
en&
Work
Alt
-Cou
ntr
yce
6d
348
9c7
d5861184642e8
b6d
f6ed
4c
Lu
shL
ife
Joe
Pas
sV
irtu
oso
No
4D
isc
1Jazz
fb1155d
c0ce
c345418f5
1d
5c8
e691d
b3
Blu
rM
ake
Do
An
dM
end
Eve
ryth
ing
You
Ev..
.P
un
ka9e3
f4d
9d
86e6
cdc8
469b
a39899d
2397
Bla
ckT
ongu
eM
asto
don
Th
eH
unte
rS
ton
erM
etal
1678231282d
d253ef
c2cf
264261645f8
Mom
ents
Pas
sH
otW
ater
Mu
sic
Moon
pie
sF
or
Mis
fits
Pu
nk
c95b
f50a1cb
a144c2
ad
40c5
211061e3
0
59
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
New
Inst
rum
enta
lH
otW
ater
Mu
sic
Nev
erE
nd
erD
emos
Pu
nk
4b
b13b
07af7
69b
a313f5
e50c5
6ee
6cf
c
Nig
ht
and
Day
Joe
Pas
sV
irtu
oso
Jazz
99d
b2a8b
625275fa
2d
9750f7
04d
078ae
Nos
talg
iaN
oM
otiv
An
dT
he
Sad
nes
sP
...
Pu
nk
9ff
1e3
d318d
3b
dae4
24d
c08e3
e6ad
30f
Ob
livio
nM
asto
don
Cra
ckth
eS
kye
Sto
ner
Met
al
6eb
21ab
c6f4
629f9
79eb
43b
b115ec
5e5
Orc
hes
trio
nP
atM
eth
eny
Orc
hes
trio
nJazz
54c5
a526e4
6a85f6
e04388171d
d50f0
a
Pap
’sB
lues
Her
bE
llis
Noth
ing
Bu
tth
eB
...
Jazz
7a0332c6
2fd
406519954646cf
eb34c9
1
Pla
net
Of
Sou
nd
Pix
ies
Dea
thT
oT
he
Pix
ies
Alt
ern
ati
ve05d
37c2
252c9
758e2
0aab
bfc
5e2
81247
Pla
yM
ike
Ste
rnP
lay
Jazz
770ab
c304d
d540713b
dc3
0c3
7270ac1
9
Poi
son
Hot
Wat
erM
usi
cT
he
New
Wh
at
Nex
tP
un
k9d
bb
d1b
46b
6013f3
2b423dc0
566c2
7d
d
Rem
edy
Hot
Wat
erM
usi
cC
au
tion
Pu
nk
ba4956d
ba1a0673a719c1
0422e6
7351a
Sh
ine
Her
bE
llis
Th
eC
on
cord
Jazz
...
Jazz
fc34c9
cae4
4c6
06a7ee
3f0
6e9
caa7662
Son
ghou
seK
evin
Eu
ban
ks
Sh
ad
owP
rop
het
sJazz
bd
04468e3
ba5d
960554d
bec
3682d2ad
8
Sou
thE
ast
Fir
stH
otW
ater
Mu
sic
No
Div
isio
nP
un
ka629874a67ce
03424cd
98b
ea84584c3
f
Sto
p!
Jan
e’s
Ad
dic
tion
Rit
ual
de
loh
ab
i...
Alt
ern
ati
ved
172248
6ee
7240b
9e3
9369652f8
e5c4
4
Tec
hn
oJoh
nS
cofi
eld
Sti
llW
arm
Jazz
1a78842d
a6cc
94b
4f1
65320b
e1b
c89f5
Tel
lY
ouW
hat
Joh
nS
cofi
eld
Lou
dJazz
Jazz
2760cc
66254b
942704962d
9d
3d
676301
Tem
pta
tion
Wai
tsG
arb
age
Ver
sion
2.0
Alt
ern
ati
ve91fd
2ca
ba6f4
8f0
24b
7b
b31ea
20776e1
Th
eP
hin
eas
Tra
ne
Pat
Mar
tin
oT
hin
kT
an
kJazz
70d
d674ca
3a11607352e8
06d
d2f5
39c9
Th
eS
oun
dO
fS
ilen
ceP
atM
eth
eny
Wh
at’
sIt
All
Ab
ou
tJazz
4097e5
a9b
8ef
6757d
2b
48e4
1028f4
977
Th
eS
tory
Tel
ler
Kev
inE
ub
anks
Th
eS
earc
her
Jazz
3502f7
daff
6115b
5d
167c4
52a84521b
0
60
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Th
eV
isit
Pat
Mar
tin
oF
ootp
rints
Jazz
898911e0
8e2
fbf2
031d
3220cc
eb2d
6b
b
Th
eW
olf
IsL
oos
eM
asto
don
Blo
od
Mou
nta
inS
ton
erM
etal
d39d
268d
1543a14401cb
f6d
bc0
3d
2b
3f
Ton
y’s
Gir
lL
aym
enT
erm
sA
nIn
trod
uct
ion
Ind
iee2
9b
82b
6ef
dcf
4e7
a1cd
e5e2
6171e5
9f
Tra
nsl
oca
tion
Hot
Wat
erM
usi
cF
ore
ver
an
dC
ounti
ng
Pu
nk
5c9
83b
b4488ed
47828866512c0
2ee
a14
Tu
mb
leH
ome
Mik
eS
tern
Wh
oL
etT
he
Cats
...
Jazz
eb1e6
87754b
7d
50ab
596fa
a8b
02504b
c
Tu
rnin
gP
oint
(Par
t1)
Kev
inE
ub
anks
Tu
rnin
gP
oin
tJazz
ac1
c02a1d
358659d
2d
f8b
821a7907a68
Wab
ash
III
Joh
nS
cofi
eld
Tim
eon
My
Han
ds
Jazz
b9cf
0e1
97e5
ecc6
d527b
7069e6
6a447a
Wal
k,
Don
’tR
un
!Joh
nny
Sm
ith
Walk
,D
on
’tR
un
Jazz
6a630423b
6c4
04f1
4b
911d
c318fd
f604
Wh
ere
Or
Wh
enJoh
nny
Sm
ith
Moon
light
InV
erm
ont
Jazz
9e8
648757e5
3fa
b0b
efe2
a4f3
83315b
6
Wil
ma’
sR
ainb
owH
elm
etB
etty
Alt
ern
ati
vec7
69711927c2
f38057c5
811a90a48587
Ble
wN
irva
na
Ble
ach
Alt
ern
ati
veb
84d
f3fe
188586c7
9804cb
068d
67a373
Bod
omA
fter
Mid
nig
ht
Ch
ild
ren
ofB
od
om
Foll
owT
he
Rea
per
...
Met
al
98d
78b
1f0
9b
a0435ef
1ad
a4ec
d955260
AM
ovie
Scr
ipt
En
din
gD
eath
Cab
for
Cu
tie
Th
eP
hoto
Alb
um
Ind
ied
550fa
d4b
a71e4
c04c9
8a7e9
e30b
1753
Act
Ap
pal
led
Cir
caS
urv
ive
Ju
turn
aIn
die
8e2
8361c1
b1b
a9fc
7ec
97b
c837508e1
6
Alb
atro
ssC
orro
sion
Of
Con
for.
..D
eliv
eran
ceS
ton
erM
etal
7103d
feed
579b
8a2a65710e9
2960e3
a2
All
Ner
eid
sB
ewar
eC
hio
dos
All
’sW
ell
Th
at
E..
.P
un
k3ad
2d
bcd
b85cc
28139a98346eb
b44ad
f
All
the
Th
ings
You
Are
Ch
arli
eP
arker
Yard
bir
dS
uit
eD
i...
Jazz
9038e5
6d
51702440fe
704aeb
fcc0
1fe
5
Alw
ays
&N
ever
Coh
eed
and
Cam
bri
aG
ood
Ap
oll
o,
I’m
...
Pu
nk
886b
6fd
96e0
4ab
330529a4c3
ef4c3
d28
Am
eric
anG
irls
Cou
nti
ng
Cro
ws
Hard
Can
dy
Alt
ern
ati
ve7c3
6a3ce
1e0
4760b
4b
ad
18b
b3ab
41ae7
An
gels
ofth
esi
len
ces
Cou
nti
ng
Cro
ws
Rec
over
ing
Th
eS
a..
.A
lter
nati
ve68af5
f9408658ec
12432a15737f7
842e
61
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
An
oth
erD
ayD
ream
Th
eate
rIm
ages
And
Word
sP
rog.
Met
al
dd
9709a7b
778c8
785fc
14b
00b
5088c6
6
Are
You
Dea
dY
et?
Ch
ild
ren
ofB
od
om
Are
You
Dea
dY
et?
Met
al
5b
1a67e
73d
68d
39ad
3b
1d
4a1906b
d9b
b
Bad
Boy
(Ori
gin
alm
ix)
Eri
cC
lap
ton
Eri
cC
lap
ton
Del
u..
.B
lues
eeb
eecb
6f3
a0765ec
7179098d
a1d
5909
Bad
Boy
Eri
cC
lap
ton
Eri
cC
lap
ton
Del
u..
.B
lues
b823ff
29520d
91e1
b18e2
45010d
17b
ee
Bef
ore
You
Acc
use
Me
Eri
cC
lap
ton
Un
plu
gged
Blu
es7cf
a5cf
21363336d
bf2
159128e3
1a3a6
Ben
son
Hed
ges
Fu
n.
Aim
an
dIg
nit
eIn
die
745f7
989345ac5
8a3a7842af5
f52ff
ed
Cal
idos
cop
ico
Diz
zyG
ille
spie
Afr
o-C
ub
an
Jazz
M..
.Jazz
83d
5d
850183356a0287b
88019747a4fd
Cau
ght
InA
Web
Dre
amT
hea
ter
Aw
ake
Pro
g.
Met
al
29e3
c87a4c1
7f0
8fb
5cd
4d
47b
1a5c7
1c
Ch
oos
eth
eO
ne
Wh
o..
.C
opel
and
InM
oti
on
Ind
ie908f6
e6667b
894cd
0cc
b1d
b25f1
1c0
d4
Dye
dO
n8
Ch
rist
ieF
ront
Dri
ve
Anth
olo
gy
Ind
ied
5c6
20b
9e7
b4347f2
b11ae1
87b
aaf4
81
Cod
esan
dK
eys
Dea
thC
abfo
rC
uti
eC
od
esan
dK
eys
Ind
iea8c2
f618896cf
a1fe
6230660ee
6e4
f26
Com
eB
ack
Hom
eP
ete
Yor
nD
ayI
Forg
ot
Alt
ern
ati
veb
a63d
bb
1c9
b87c9
d3cb
5a27280288d
6f
Com
fort
able
Lia
rC
hev
elle
Won
der
Wh
at’
sN
ext
Met
al
c38350e2
f012ef
1f5
915b
420b
02e5
ad
6
Tes
tin
gT
he
Str
ong
...
Cop
elan
dB
enea
thM
edic
ine
...
Ind
ie2b
c8c7
b34ef
6ab
48016525e2
d7ec
c679
Dam
ned
For
All
Tim
eC
orro
sion
Of
Con
for.
..B
lin
dS
ton
erM
etal
d66203b
b71b
79af0
48822d
94d
43160c4
Fly
On
Th
eW
ind
scre
...
Dep
ech
eM
od
eB
lack
Cel
ebra
tion
New
Wav
e19b
150f
b6eb
e6b
07e5
c4502d
c94fb
04c
Sh
ine
Dep
ech
eM
od
eE
xci
ter
New
Wav
efb
02657
3d
1e5
3273a589cb
a2459a4b
0f
Th
eL
ove
Th
ieve
sD
epec
he
Mod
eU
ltra
New
Wav
e028960d
2ef
a586e8
e8b
407a24b
b809ed
Wal
kin
gin
My
Sh
oes
Dep
ech
eM
od
eS
on
gs
of
Fait
han
...
New
Wav
e416b
a82
23d
fe81950369aec
3b
638e8
e1
Din
g-D
ong
Dad
dy
of..
.C
her
ryP
opp
in’
Dad
die
sZ
oot
Su
itR
iot
Sw
ing
00343b
8fc
fa1ea
f306f8
d711c5
cc4d
83
62
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Cat
sin
aB
owl
Din
osau
rJr.
Din
osa
ur
Alt
ern
ati
ve9b
ca35b
c7b
8317a705c7
d8919ec
6a64a
Ina
Jar
Din
osau
rJr.
Foss
ils
Alt
ern
ati
vec1
372829957d
c7fe
2ce
48f9
55b
b201b
e
No
Bon
esD
inos
aur
Jr.
Bu
gA
lter
nati
vefd
4a32f
cc83d
3248ef
1ce
755b
f27d
f4e
Sid
eway
sD
inos
aur
Jr.
Wh
ate
ver’
sC
ool
W..
.A
lter
nati
ve70fe
3a75c5
cd85b
88e7
1a8ec
f861fc
da
Wh
atE
lse
IsN
ew[L
...
Din
osau
rJr.
ID
on
’tT
hin
kS
oA
lter
nati
ve898cc
f7b
6709365f2
0d
c51ee
7d
5b
9d
23
God
Wil
lin
gD
rop
kic
kM
urp
hys
Th
eM
ean
est
Of
Ti.
..P
un
k2e0
4c8
d1af0
e52c2
2b
70c0
8219f7
e37b
Du
de,
you
hav
en
oi.
..H
arri
son
Ber
ger
on
Harr
ison
Ber
ger
on
Pu
nk
9315a9651b
a7e6
042d
10d
f6ef
eeb
78fa
Eve
nF
low
Pea
rlJam
Ten
(Del
uxe
Ed
itio
n)
Alt
ern
ati
ve08b
621e
9480b
f6d
2275b
7c8
22098ae1
4
Th
eO
nly
Mon
ent
We
...
Exp
losi
ons
InT
he
Sky
Th
eE
art
hIs
Not
...
Ind
ie0502b
fc75ff
d552f1
dd
360d
398a66c2
7
Wel
com
e,G
hos
tsE
xp
losi
ons
InT
he
Sky
All
of
aS
ud
den
I...
Ind
ie8f8
78a5d
e9aa063f1
844ed
7f3
6d
68027
I’m
Not
Afr
aid
Fac
eT
oF
ace
Don
’tT
urn
Aw
ayP
un
k9025eb
6599ef
a5d
994b
d219959ef
b85a
Pri
de
War
Fu
rth
erS
eem
sF
ore
ver
Hop
eT
his
Fin
ds
Y..
.In
die
5b
d012b
32e6
7f4
49aab
d762456f0
1ed
a
Fu
ryof
the
Sto
rmD
rago
nF
orce
Son
icF
ires
torm
Met
al
85642501f7
6927f4
d7215856d
bf2
7a44
Hap
py
En
din
gsC
ounte
rfit
Su
per
Am
use
men
tM
...
Ind
ieac5
9e9
f2a12177f1
2143c3
cc75f9
44e0
Hid
eN
oth
ing
Fu
rth
erS
eem
sF
ore
ver
Hid
eN
oth
ing
Ind
ie23d
a715
407d
c4a61d
30122d
f6986f8
b4
Hol
eT
oF
eed
Dep
ech
eM
od
eS
ou
nd
sO
fT
he
Un
i...
New
Wav
e17fb
0f9
21a438a56b
c946c2
bae1
9cb
ed
IC
an’t
Get
Sta
rted
Diz
zyG
ille
spie
Diz
zyA
tmosp
her
eJazz
5f7
df3
2641488f3
5e9
b209ee
18a0307a
IF
eel
Bet
ter
Fri
ghte
ned
Rab
bit
Th
eM
idn
ight
Org
a..
.In
die
78186b
cfcf
11d
7b
a84c8
68a1d
bee
7f2
2
IW
asB
orn
InT
he
Dar
kF
irew
orks
Gosp
elP
un
kd
b3ed
a2c3
07024b
c54196f2
8d
2fa
b13c
InM
yP
lace
Col
dp
lay
AR
ush
Of
Blo
od
T..
.In
die
ae1
17ec
4d
5ae5
c3ff
6f3
926e1
976681a
63
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
So
Wh
atE
lse
IsN
owJ
Mas
cis
Mart
in+
Me
Alt
ern
ati
ve432b
eff8d
c41987ce
f1ad
8af3
7a6023c
Joh
nth
eR
evel
ator
Dep
ech
eM
od
eP
layin
gth
eA
ngel
New
Wav
e62a8d
9877b
c0fe
258f4
571d
8f1
513c1
a
Kil
lT
he
Mu
sic
Eve
ryT
ime
ID
ieG
utt
erP
hen
om
enon
Met
al
e60a5705d
6f2
e84fd
46cd
12f1
a76b
bd
b
Les
son
sD
amie
raM
US
ICIn
die
ba640d
2cb
1fd
4d
20e3
798b
5c8
1cd
d775
Lig
htn
ess
Dea
thC
abfo
rC
uti
eT
ran
satl
anti
cism
Ind
ied
b35716ff
e8fe
f1709b
bd
2ed
cea65a8b
Lon
gW
hip
,Big
Am
eric
aC
orro
sion
Of
Con
for.
..W
iseb
lood
Sto
ner
Met
al
5b
6c2
af2b
cc1a795211a4ab
04d
b69935
Low
Foo
Fig
hte
rsO
ne
By
On
eA
lter
nati
ve9123574d
7d
5642cc
c0a7a2f6
5a6e5
2cd
Mil
lsto
ne
Bra
nd
New
Th
eD
evil
An
dG
od
...
Pu
nk
64a67b
1159aa6b
271938b
c2d
833428a6
Mrs
.P
otte
r’s
Lu
llab
yC
ounti
ng
Cro
ws
Th
isD
eser
tL
ife
Alt
ern
ati
ve8cb
aab
8e2
439d
21ce
8d
337f3
bad
e8958
Mu
sic
Mu
sic
Eye
dea
&A
bil
itie
sF
irst
Born
Hip
-Hop
f6d
f48e
5d
9ed
495a7e2
e8c0
11aec
808d
Mu
sic
Now
Fri
ghte
ned
Rab
bit
Sin
gT
he
Gre
ys
Ind
ie33d
a5a6
d77e0
903c5
492e5
f57b
51436a
My
Mel
anch
oly
Bab
yD
izzy
Gil
lesp
ieB
irk’s
Work
sJazz
68d
75f6
2c8
4a334f4
20e4
962a1728792
N.S
.U.
Cre
amT
he
Bes
tO
fC
ream
...
Rock
6d
453d
0c9
4f7
3c2
dd
686e3
7274323890
New
Yea
rC
ounte
rfit
Fro
mF
inis
hto
St.
..In
die
ae3
431d
01ad
d91b
85a913375a7526567
Nob
ody
Kn
ows
Th
eT
r...
Diz
zyG
ille
spie
Imp
rom
ptu
Jazz
f42aa40d
2c5
4c7
4d
6e3
8089b
037303fa
Not
hin
gW
ith
You
Des
cen
den
tsC
ool
To
Be
You
Pu
nk
4a83c6
e14575e3
c02758a05c8
063f5
a5
Ob
livio
nO
cean
Pai
nO
fS
alva
tion
Th
eP
ain
ful
Ch
ron
...
Pro
g.
Met
al
71b
0806
1407f7
e0f6
7f7
c6d
5c5
119213
Off
Bro
adw
ayE
very
Tim
eI
Die
Hot
Dam
n!
Met
al
f39af5
f7724f7
a35d
50afc
bfa
7866b
15
Om
aha
Cou
nti
ng
Cro
ws
Au
gu
stan
dE
very
t...
Alt
ern
ati
ve0683ce
0a8e6
1aa411b
cf634ad
a78e5
e4
Ou
tof
Now
her
eC
har
lie
Par
ker
Yard
bir
dS
uit
eD
i...
Jazz
e9aca
351c6
4430e9
ab
5639508b
5d
4b
55
64
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Ove
rtu
re19
28D
ream
Th
eate
rM
etro
poli
sP
t.2
S..
.P
rog.
Met
al
ce125c9
6fd
2079b
3114d
f6778ef
18802
Par
anoi
dO
pio
idC
orro
sion
Of
Con
for.
..In
Th
eA
rms
Of
God
Sto
ner
Met
al
b686fe
5d
2b
1e2
93b
dd
599969cd
3139e7
Pre
sid
ent
ofW
hat
?D
eath
Cab
for
Cu
tie
Som
eth
ing
Ab
ou
tA
...
Ind
iee6
ec56fb
d38c7
cc3ca
a8d
ebd
f7c4
bb
7d
Pri
son
erD
okke
nB
ack
For
Th
eA
ttack
Met
al
9b
d0af2
39262b
f538e9
3c9
9d
ebc2
9b84
Ray
’sId
eaD
izzy
Gil
lesp
ieC
ool
Bre
eze
Jazz
ca07d
37ee
5ff
89c4
8054e2
e7b
103971f
Red
IsT
he
New
Bla
ckF
un
eral
For
AF
rien
dS
even
Way
sT
oS
cr..
.P
un
kfb
4179a
64e7
497e3
d6d
eb9997b
907b
c2
Rock
etD
efL
epp
ard
Hyst
eria
Met
al
c18cc
3c9
b59869294aacc
13f1
64c0
070
Sal
tP
eanu
tsD
izzy
Gil
lesp
ieS
alt
Pea
nu
tsJazz
190c3
7e0
431922fe
68aa56a70e6
198c5
Say
ItT
oM
yF
ace
Ch
erry
Pop
pin
’D
ad
die
sK
ids
on
the
Str
eet
Sw
ing
0e8
35ae0
6b
34e1
67a4f4
ed7692e8
235f
Scr
eam
ing
Infi
del
itie
sD
ashb
oard
Con
fess
ion
al
Th
eP
lace
sY
ou
Ha..
.In
die
c58e0
9ab
64623b
24fe
a3485b
daca
e60f
Sh
ine
Fad
edG
rey
AQ
uie
tT
ime
Of
D..
.P
un
k38d
332f
a0c3
c7ee
4b
d6626463acd
34b
6
Six
pou
nd
erC
hil
dre
nof
Bod
om
Hate
Cre
wD
eath
roll
Met
al
3071674b
36b
851a7fc
490fc
66a8c8
055
Sn
owfa
llE
mil
yR
emle
rE
ast
To
Wes
Jazz
0e3
ab
2980763afe
8a23aff
bd
d14b
6387
Sock
able
Fac
eC
lub
Ch
erry
Pop
pin
’D
ad
die
sR
ap
idC
ity
Mu
scle
...
Sw
ing
c292749478c4
2e4
96e7
5cf
746489cd
0e
Som
eN
ights
Fu
n.
Som
eN
ights
Ind
ie551864b
c887cc
3b
2664599166c0
f70b
e
Sou
nd
trac
kT
oT
he
Rid
eF
airw
eath
erIf
Th
eyM
ove.
..K
i...
Pu
nk
2c0
a22d
04893338f2
343ef
e75200d
6ea
Sp
ook
yG
irlf
rien
dE
lvis
Cos
tell
oW
hen
IW
as
Cru
elR
ock
7074c7
5ed
658eb
53325176b
e532ce
29c
Sw
imU
nti
lY
ouC
an’.
..F
righ
ten
edR
ab
bit
Th
eW
inte
rof
Mix
...
Ind
ie87fc
ae8
4ef
cef8
0e6
5eb
ef7c2
62d
50c4
Sw
ingi
ng
Wit
hT
iger
...
Ch
erry
Pop
pin
’D
ad
die
sS
ou
ldC
ad
dy
Sw
ing
bc0
41fe
323b
7108d
8c3
a3b
ecca
6f2
c30
Tak
eth
e”A
”T
rain
Du
keE
llin
gton
Bes
tof
Du
keE
lli.
..Jazz
ff3d
4f4
153d
ffd
b6a0c4
bee
4aa4b
b833
65
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Tea
rsB
efor
eB
edti
me
Elv
isC
oste
llo
Imp
eria
lB
edro
om
Rock
60a175a366c5
89b
a2f4
6ef
5ce
689669a
Tee
nag
eB
rain
Su
rgeo
nC
her
ryP
opp
in’
Dad
die
sF
eroci
ousl
yS
ton
edS
win
g320629ab
9e9
c7cb
3d
580fb
d97a7c5
cff
Th
eA
nsw
erL
ies
Wit
hin
Dre
amT
hea
ter
Oct
avari
um
Pro
g.
Met
al
99a8139b
fb61fe
6a63afd
bd
e3b
b6bb
28
Th
eB
oat
Chu
ckR
agan
Los
Fel
izIn
die
20b
5378
aa8b
2f8
23e6
5e7
83d
e129e3
10
Th
eB
rad
ley
Fu
rth
erS
eem
sF
ore
ver
Th
eM
oon
IsD
own
Ind
ief9
4107fa
55b
383e9
e1e4
4b
e2ac4
14cb
e
Th
eE
mp
loym
ent
Pag
esD
eath
Cab
for
Cu
tie
We
Hav
eT
he
Fact
s...
Ind
ie708ae6
11b
0116f9
75e8
367f1
af4
0f6
3b
Th
eH
unte
rD
okke
nU
nd
erL
ock
An
dK
eyM
etal
8ea
7a84b
0c5
d805970a75180e9
4f9
6d
6
Th
eS
oun
dF
urt
her
See
ms
Fore
ver
How
To
Sta
rtA
Fir
eIn
die
460813d
3c9
bb
a12ca
4b
5983612542e7
e
Th
eT
ake-
Aw
ayF
ace
To
Fac
eH
owT
oR
uin
Eve
ry..
.P
un
k5121493cf
260497c7
16a8601acf
282ae
Th
isY
ear’
sG
irl
Elv
isC
oste
llo
Th
isY
ears
Mod
elR
ock
5c8
75951fe
5b
c831c8
8474eb
81d
2afd
a
Val
ley
Of
Th
eD
amn
edD
rago
nF
orce
Vall
eyO
fT
he
Dam
ned
Met
al
523b
db
757b
eb279e1
e65c9
fba1147648
Wai
tin
gF
orA
Gir
l..
.F
orei
gner
Th
eD
efin
itiv
eC
o..
.R
ock
5b
fb480f6
50d
fbfa
0fe
f2f1
034a60c7
2
Yas
min
Th
eL
ight
Exp
losi
ons
InT
he
Sky
Th
ose
Wh
oT
ell
Th
...
Ind
ie5fd
43ae
c7204b
09ec
7d
3194fa
deb
4f8
b
You
rH
and
InM
ine
(...
Exp
losi
ons
InT
he
Sky
Fri
day
Nig
ht
Lig
hts
Ind
ied
7d
111769c3
3cc
4d
793240cd
7c6
8725c
An
dro
gyny
Gar
bag
eb
eau
tifu
lgarb
age
Alt
ern
ati
ve50ad
a5a
80f7
85c6
4c2
95cc
e23d
d7311d
Bac
kyar
dS
ku
lls
Fri
ghte
ned
Rab
bit
Ped
estr
ian
Ver
seIn
die
5221ac8
710765d
bb
549c3
2866aca
8a4c
Blo
od
dru
nk
Ch
ild
ren
ofB
od
om
Blo
od
dru
nk
Met
al
b42f9
3347e6
d9ed
6c2
53235e9
1814d
d0
Cem
eter
ies
Of
Lon
don
Col
dp
lay
Viv
aL
aV
ida
Ind
ie0ae0
0c2
6861a1af4
ed5e9
df7
80910d
a6
AD
iam
ond
and
aT
eth
erD
eath
Cab
for
Cu
tie
Th
eO
pen
Door
EP
Ind
ie2b
29540
4f9
bb
d96fe
7f7
4093cd
307b
53
IW
ill
Pos
sess
You
r...
Dea
thC
abfo
rC
uti
eN
arr
owS
tair
sIn
die
4a94eb
c76eb
96b
85a611ea
d94b
0d
42d
5
66
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Th
eT
hin
gsY
ouS
aid
Dep
ech
eM
od
eM
usi
cF
or
Th
eM
ass
esN
ewW
ave
2526969d
5b
7f7
f50b
36b
d52a18a9b
a65
Now
Eye
dea
&A
bil
itie
sE
&A
Hip
-Hop
1168eb
29fa
76f4
49d
8d
7b
25243370b
24
Sp
inC
ycl
eE
yed
ea&
Ab
ilit
ies
By
Th
eT
hro
at
Hip
-Hop
95b
bee
bcb
13164815aec
c66b
01592fb
7
Goi
ng
Ou
tIn
Sty
le..
.D
rop
kic
kM
urp
hys
Goin
gO
ut
InS
tyle
Pu
nk
23579b
0c2
ba9137cc
b8531872a8d
e66c
Hu
man
Qu
alit
ies
Exp
losi
ons
InT
he
Sky
Take
Care
,T
ake
C..
.In
die
5825a8b
04754d
ba5b
6449ed
d295a1472
IS
omet
ime
Wis
hI
W..
.D
epec
he
Mod
eS
pea
k&
Sp
ell
New
Wav
e4599580ca
86ac9
640223f7
b8215013ef
It’s
Not
All
Ab
out
You
Fac
eT
oF
ace
Lau
gh
Now
...L
au
gh
...
Pu
nk
dd
2001395a74223e7
427fc
d6b
4ff
402a
Mon
um
ent
Dep
ech
eM
od
eA
Bro
ken
Fra
me
New
Wav
e7f5
a80298d
954d
29604b
ccafb
3081c9
f
Mor
eT
han
aP
arty
Dep
ech
eM
od
eC
on
stru
ctio
nT
ime.
..N
ewW
ave
1f5
b29c
d5ac4
c1b
9f8
ba263ce
a23f6
22
Rad
ioC
hri
stie
Fro
nt
Dri
ve
Ch
rist
ieF
ront
Dri
ve
Ind
ie8127c2
a615d
e6141e7
77a2859400a280
Ru
nB
aby
Ru
nG
arb
age
Ble
edL
ike
Me
Alt
ern
ati
ve401fa
67cd
2611257ee
fccf
8f8
516fb
64
Sh
owM
eM
ary
(XF
MR
...
Cat
her
ine
Wh
eel
Way
dow
n[1
]A
lter
nati
ve29d
2236
4f4
485e9
5a1a4d
671f9
6ed
9f6
Sn
owA
nd
Lig
hts
Exp
losi
ons
InT
he
Sky
How
Str
an
ge,
Inn
o..
.In
die
ea741ea
0f6
1a0ed
f07c0
3d
d9b
ee6107b
Sta
rtA
gain
Cou
nti
ng
Cro
ws
Un
der
wate
rS
un
shin
eA
lter
nati
ve5366f1
dc9
2c0
fe2ec
8d
07d
6f2
28b
e3fb
Why
You
’dW
ant
To
L..
.D
eath
Cab
for
Cu
tie
Th
eJoh
nB
yrd
E.P
.In
die
a9622b
8e5
cd3ef
e507fe
af4
1cc
e3e0
9d
3C
hor
ds
Con
stel
lati
onof
Cars
Con
stel
lati
on
of
...
Ind
ie1d
7e3
f4c8
0b
264e8
6b
20d
97f1
38d
0953
Blu
e’n
’B
oog
ieD
izzy
Gil
lesp
ieJazz
Coll
ecto
rE
d..
.Jazz
bd
922ee
fbce
13a2993cd
6e3
1ca
0d
c114
Hu
rts
Lik
eH
eave
nC
old
pla
yM
ylo
Xylo
toIn
die
81179f4
fccb
606997f2
23937e8
6143f5
Th
eG
rey
Man
Cop
elan
dY
ou
Are
My
Su
nsh
ine
Ind
ie77a0e7
cb790f6
2e6
47d
bc8
aac8
14ce
e3
Tec
hn
icol
orG
irls
Dea
thC
abfo
rC
uti
eF
orb
idd
enL
ove
E.P
.In
die
ed7967e
573592a991834b
232f2
b550b
e
67
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Pu
ke+
Cry
Din
osau
rJr
Gre
enM
ind
Alt
ern
ati
vece
b2613
e2d
3627b
a65a7e2
57a152223c
Eve
ryth
ing
toE
very
one
Eve
rcle
arS
oM
uch
for
the
A..
.A
lter
nati
ve9fc
ae8
756a683b
eb88a14c2
f57d
44086
IK
now
You
Wel
lF
ace
To
Fac
eB
igC
hoic
eP
un
ka7ab
7a6
844c3
209d
308e0
0d
c2d
4043a5
IW
ant
Fac
eT
oF
ace
Liv
eP
un
kd
592c3
961690102a4429666957c6
6780
InH
arm
sW
ayF
ace
To
Fac
eIg
nora
nce
isB
liss
Pu
nk
8cf
60f5
26939b
3c8
93ee
7ad
10b
695e1
b
Ou
tof
Focu
sF
ace
To
Fac
eR
eact
ion
ary
Pu
nk
1f0
21878164d
df3
00cff
c3011762100f
Wal
kT
he
Wal
kF
ace
To
Fac
eF
ace
To
Face
Pu
nk
b4e4
f247c6
99f1
7a84ce
ebc5
94ed
04b
a
Wel
com
eB
ack
To
Not
...
Fac
eT
oF
ace
Th
ree
Ch
ord
sA
nd
...
Pu
nk
0139690a26f6
3c3
5a2e7
5278a67b
961c
Com
eA
rou
nd
Fir
ewor
ks
All
IH
ave
To
Off
...
Pu
nk
03ec
f00e7
e0afd
5b
aa8003b
6d
623456a
Pu
mp
edU
pK
icks
Fos
ter
Th
eP
eop
leT
orc
hes
Ind
ie0f8
fd09
1754a4633e1
c20fa
9a9474ce
7
Res
cue
Tra
ined
Fu
rth
erS
eem
sF
ore
ver
Pen
ny
Bla
ckIn
die
68b
256d
c0c9
65f4
c10c5
fbaa867ef
9fd
Gas
olin
eC
ath
erin
eW
hee
lW
ishvil
leA
lter
nati
ve5f7
852412d
8ce
cdd
19fc
8ef
6c6
1e5
de9
He
Lov
esan
dSh
eL
oves
Ell
aF
itzg
erald
an
d...
Ell
a&
Lou
isS
ing..
.Jazz
22905ef
5a9e4
e305d
60b
0b
8e3
c7e4
f87
Her
eW
eG
oA
gain
Eve
rcle
arS
on
gs
Fro
man
Am
e...
Alt
ern
ati
vee5
3b
6d
a8a35c0
b1b
201725a932657265
Her
oin
Gir
lE
verc
lear
Sp
ark
lean
dF
ad
eA
lter
nati
ved
f5ca
a3f7
fc96e0
d03270f9
048127147
How
Inse
nsi
tive
Em
ily
Rem
ler
Ret
rosp
ecti
veV
ol.
..Jazz
e85d
221
75cc
217a0103b
0b
efc6
5d
82a8
InK
eep
ing
Sec
rets
...
Coh
eed
and
Cam
bri
aIn
Kee
pin
gS
ecre
t...
Pu
nk
5f5
50e6
ee34277d
0b
06059af4
9879ab
4
Lov
eIs
Her
eT
oS
tay
Ell
aF
itzg
erald
an
d...
Th
eB
est
Of
Ell
a..
.Jazz
19ee
dd
781e4
9085108b
e3e6
886fa
a40f
No
Wor
ldfo
rT
omor
row
Coh
eed
and
Cam
bri
aG
ood
Ap
oll
oI’
mB
...
Pu
nk
35ca
c2568267a7fd
48d
286c8
77a47d
10
Nu
nca
Mai
sE
mil
yR
emle
rR
etro
spec
tive
Vol.
..Jazz
1536a82cb
cd217576ef
4cc
42cf
45d
53a
68
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Qu
eer
Gar
bag
eG
arb
age
Alt
ern
ati
vee0
967550e0
b17837717e1
7ef
37a5b
999
Sic
kan
dT
ired
Eve
rcle
arW
orl
dO
fN
ois
eA
lter
nati
ved
672cb
1b
b11709b
c0c9
5163d
877290e0
Th
eJit
terb
ug
Wal
tzF
ats
Wal
ler
Port
rait
Vol.
2-
...
Jazz
cf8244f7
bab
5f5
2ae5
fe3aa57fe
5e6
35
Wh
enD
ayIs
Don
eD
jan
goR
ein
hard
tT
he
Bes
tof
Dja
ng..
.Jazz
ded
ff1fd
9fe
af9
3419e4
8ee
1326104ff
Wh
olly
Cat
sC
har
lie
Ch
rist
ian
Th
eO
rigin
al
Gu
it..
.Jazz
6b
28142
d120417034e9
2651b
442f2
f58
Wh
atIf
Col
dp
lay
X&
YIn
die
6b
4a699
74a92109ad
762252e6
9cc
60ac
Sh
ovel
Kn
ock
out
Ch
ild
ren
ofB
od
om
Rel
entl
ess,
Rec
kl.
..M
etal
f803c7
3f1
0d
75e6
a10779d
84b
b19fe
5c
At
the
Bot
tom
Bra
nd
New
Dais
yP
un
kfb
991d
503ff
1450b
962c7
5d
325d
ecec
8
B-B
oys
Mak
in’
Wit
h..
.B
east
ieB
oys
Ill
Com
mu
nic
ati
on
Hip
-Hop
fbd
5b
25c6
ea3f1
f33ff
dcc
c39a0ff
f1c
Bes
tD
ays
Blu
rT
he
Gre
at
Esc
ap
eA
lter
nati
ve94d
4d
39cc
7886b
7d
c3fc
92970318285b
Bla
ckM
etal
lic
Cat
her
ine
Wh
eel
Fer
men
tA
lter
nati
veb
ccc9
59a8202fd
3017e0
447d
eafd
d90f
Sta
nd
ing
InT
he
Doo.
..B
obD
yla
nT
ime
Ou
tO
fM
ind
Folk
2afe
5e2
ec48aae2
0b
52349e4
45564f3
0
Bom
bB
ush
Six
teen
Sto
ne
Alt
ern
ati
ve1f4
5c9
a12a9f7
33c5
a5e1
26d
d5cc
b350
Eu
lali
a,E
ula
lia
Bra
idM
ovie
Mu
sic
vol.
Tw
oIn
die
742ee
5ff
71d
b928af3
8d
6d
9671452e0
7
Min
uet
Bra
idF
ran
kie
Wel
fare
B..
.In
die
9ca
08d
e8c9
862a58d
423b
d93b
51d
64f2
Bri
ckB
enF
old
sF
ive
Wh
ate
ver
An
dE
ver.
..In
die
be7
9c1
95d
bb
7f3
538cf
e0364f6
d576c6
bro
ken
dow
np
iece
o...
Bri
anS
etze
r13
Rock
abil
ly5f7
e94f1
fb50b
8599a224543fc
1e9
703
Cou
ntr
yS
adB
alla
dM
anB
lur
Blu
rA
lter
nati
ve0eb
7ad
6ca
06691a285e7
7b
58cd
20ea
2d
Cra
nk
Cat
her
ine
Wh
eel
Ch
rom
eA
lter
nati
vee2
b68c3
8e2
868ce
c31ed
dd
088f1
12771
Del
icio
us
Cat
her
ine
Wh
eel
Ad
am
An
dE
ve-
Me.
..A
lter
nati
ve14218fe
31d
c28ae9
b9410f4
50fa
91e4
6
69
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Slu
dge
feas
tD
inos
aur
Jr.
You
’re
Liv
ing
All
...
Alt
ern
ati
veb
3248a7
460b
b32f8
521b
8175ce
9c4
6a6
Dru
nke
nK
nig
ht
Big
wig
Un
-Mer
ryM
elod
ies
Pu
nk
30f0
af0
9a8364ca
aee
cb4101c5
731211
God
Res
tye
Mer
ryG
...
Bin
gC
rosb
yF
rank
...
Ch
rist
mas
Jazz
335ab
58ad
4e3
a569ce
72e0
2fc
240fa
d2
Hey
You
(Bat
chof
L..
.B
lack
Lab
elS
oci
ety
Son
icB
rew
Met
al
d494aad
b3b
bd
0b
6aef
6386797ac7
80b
2
Hey
,Joh
nny
Par
kF
oo
Fig
hte
rsC
olo
ur
&T
he
Sh
ap
eA
lter
nati
ve2ab
152e
36e0
fc14d
1901625603ace
fbc
Ju
stif
yth
eT
hri
llB
lues
Tra
vel
erS
traig
ht
on
Til
l..
.A
lter
nati
ve036b
eeb
2ea
5b
931d
43c1
4c7
9a226f1
38
Let
ter
Fro
mA
Fri
end
Blu
esT
ravel
erS
ave
His
Sou
lA
lter
nati
ve247c3
a344c2
7b
afa
e64a0d
0aa7fe
f107
Lif
etim
e,A
Bet
ter
Th
anE
zra
Bef
ore
Th
eR
ob
ots
Alt
ern
ati
ve574a42658e7
b235d
bb
f242a66ab
ab
7e5
Lit
tle
Mu
scle
Cat
her
ine
Wh
eel
Hap
py
Day
sA
lter
nati
ve091f6
d9c4
7388a9e4
74302d
70b
027f1
1
Lon
gL
ost
Bet
ter
Th
anE
zra
Fri
ctio
n,
Baby
Alt
ern
ati
ve97676a90d
9b
070189cc
1c2
e2b
8e4
ae3
2
Look
Aro
un
dB
lues
Tra
vel
erF
ou
rA
lter
nati
ve943e7
f1387b
4b
47246462b
ab
1a9482fa
Los
t,N
ight
Bil
lF
rise
llD
isfa
rmer
(Sta
nd
ard
)Jazz
87a4069119ab
91ad
5b
ba07e7
79e9
12b
9
Mes
sB
enF
old
sF
ive
Th
eU
nau
thori
zed
...
Ind
ie1c7
390b
a840644988f5
0a238b
0557301
Mor
eT
han
AF
eeli
ng
Bos
ton
Gre
ate
stH
its
Rock
aa1a5078ce
9c0
a595d
1ab
72e6
151ec
d4
Per
fect
Pit
chB
raid
Mov
ieM
usi
cV
ol.
On
eIn
die
11fc
7661e4
6e0
c9c2
a0a3d
d71b
b76224
Sh
e’s
Cra
fty
Bea
stie
Boy
sL
icen
sed
To
Ill
Hip
-Hop
c3d
751d
e1182d
d7ab
222536ef
b8466d
d
Sig
nO
fL
ife
Bil
lF
rise
llan
d85..
.S
ign
Of
Lif
eJazz
4b
a241c
9b
f07574ad
e5962ef
a59d
fbfc
Sm
ile
Big
wig
Sta
yA
slee
pP
un
k018a656b
5e2
0a7aaf3
5ac7
82b
aea
b139
Sou
ther
nG
rlB
ette
rT
han
Ezr
aD
elu
xe
Alt
ern
ati
ve4830e3
b50e5
2b
15d
223b
dd
c284e1
4605
Ste
pp
inS
ton
eB
lack
Lab
elS
oci
ety
Han
gov
erM
usi
cV
o..
.M
etal
149e0
45fd
7870f7
c029e6
08fe
341d
d8f
70
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Su
icid
eM
essi
ahB
lack
Lab
elS
oci
ety
Mafi
aM
etal
1fc
82f2
1c0
2aa19cc
e1d
72666fd
c87a2
Th
eN
inja
Cac
oph
ony
Sp
eed
Met
al
Sym
ph
ony
Met
al
acb
2e5
f7f5
8cc
3180315560aeb
8803d
b
Bac
kw
ards
Gu
itar
Cat
her
ine
Wh
eel
Ju
dy
Sta
rin
gA
tT
...
Alt
ern
ati
ve7ac8
2837d
e46e4
11e4
21e6
8ec
5b
2c9
bc
Col
lid
eosc
opic
Cat
her
ine
Wh
eel
IW
ant
To
Tou
chY
ou
Alt
ern
ati
ve7e0
21cc
b528cd
95b
5756a340c1
84b
243
Des
cen
din
gB
abe
Cat
her
ine
Wh
eel
Ma
Soli
tud
a[D
isc
1]
Alt
ern
ati
ve6b
ae2
c49d
16516cc
5b
ab
1ab
e0458e4
c5
Div
ers
Bra
idT
he
Age
of
Oct
een
Ind
ie86140658751af7
ece3
1c9
14d
e4e6
c374
Hea
l(L
ive
Xfm
)C
ath
erin
eW
hee
lB
roke
nN
ose
[Dis
c2]
Alt
ern
ati
vefa
92e8
61b
44fa
aea
e243fa
963acc
0941
Kil
lin
gA
Cam
era
Bra
idL
uck
yT
oB
eA
live
Ind
ie4b
03ec
ce71b
e25a068b
12944d
e3d
0116
Mas
ters
Of
War
Bob
Dyla
nT
he
Fre
ewh
eeli
n’
...
Folk
8ab
9c3
6e5
b549d
a62877ee
fc6a8fb
cbd
Mou
thfu
lO
fA
irC
ath
erin
eW
hee
lL
ike
Cats
An
dD
ogs
Alt
ern
ati
veaee
2773d
cb0ef
06a0783e3
de1
53970a9
Nev
erW
ill
Com
eF
orU
sB
raid
Fra
me
&C
anva
sIn
die
d864b
a86b
cbc9
7a82e3
fdff
a5fe
43d
7c
Tex
ture
(Liv
e)C
ath
erin
eW
hee
lB
roke
nN
ose
[Dis
c1]
Alt
ern
ati
ve1c9
389f4
b75547a0aec
e024e9
a38b
502
Th
eQ
uie
tT
hin
gsT
h..
.B
ran
dN
ewT
he
Holi
day
EP
Pu
nk
046fc
264f3
00d
d0a8f6
c9b
657fe
af4
94
Ton
gue
Tw
iste
dC
ath
erin
eW
hee
lC
ran
kA
lter
nati
ve7f6
30e3
afd
ea0e2
40a918514e2
b537e9
Wis
hC
ath
erin
eW
hee
lS
he’
sM
yF
rien
d(E
p)
Alt
ern
ati
ve189350a70199c0
fdb
8c6
d51ef
b0a398a
Alw
ays
Gon
na
Get
Ya
Big
Bad
Vood
oo
Dad
dy
Sav
eM
yS
ou
lS
win
ga96391251a887fc
c7fd
959124c8
7c8
02
Th
eC
all
Of
Th
eJit
...
Big
Bad
Vood
oo
Dad
dy
How
Big
Can
You
Get
Sw
ing
50f7
2f2
9371910c9
1e2
3b
af9
04211e5
c
Blu
esE
tud
eB
rian
Nov
aS
had
owO
fY
ou
rS
mil
eJazz
8e0
0b
0280412149e3
d51a47e0
974a110
En
dO
fA
Cen
tury
Blu
rP
ark
life
Alt
ern
ati
ve1d
7d
6b
309b
77805a45f8
2b
268b
ce5fd
0
You
Are
the
Rea
son
Bra
idC
lose
rT
oC
lose
dIn
die
4fd
5270
fc4797d
e8f2
30a6d
9b
8471d
53
71
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Coff
aro’
sT
hem
eB
ill
Fri
sell
wit
hD
...
Bil
lF
rise
llw
ith
...
Jazz
70ee
aa21093f7
086b
c79e2
9d
0b
602e5
8
Com
eR
ain
Or
Com
eS
...
Bil
lie
Hol
iday
Bil
lie’
sB
est
Jazz
09aae6
e14962d
ab
9b
963059f3
4ea
f7b
0
God
son
Son
gB
ill
Fri
sell
Gon
e,Ju
stL
ike
a..
.Jazz
1390c8
3974256d
8ec
499002e5
011e1
8b
Hap
py
Fee
lin
gB
arn
eyK
esse
lT
oS
win
gor
Not
t...
Jazz
d6b
9a57098ef
4220019d
617a1670d
db
5
Hap
py
Lit
tle
Son
gB
arn
eyK
esse
lS
olo
Jazz
84d
3cc
8c8
84d
6f6
39f5
424cc
f6c2
17f6
ItC
ould
Hap
pen
toY
ouB
arn
eyK
esse
lT
he
Poll
Win
ner
sJazz
57b
517a
19b
c649af8
90f3
3e0
4a5d
58a3
ItT
akes
AL
otT
oL
...
Bob
Dyla
nH
ighw
ay61
Rev
isit
edF
olk
1e2
a0775855425ea
dd
86e3
98137a1747
Kin
gO
fS
win
gB
igB
adV
ood
oo
Dad
dy
Big
Bad
Vood
oo
Dad
dy
Sw
ing
32e1
e622861c7
31d
0422eb
289ff
8580f
Mad
hou
seB
enny
Good
man
Ben
ny
Good
man
Jazz
451b
ee4d
3b
18737e0
2a2611ec
ad
7159d
Mag
gie’
sfa
rmB
obD
yla
nB
rin
gin
gIt
All
B..
.F
olk
d9e9
1d
304ea
0a2b
36d
ff34acb
9b
801d
9
Moz
amb
iqu
eB
obD
yla
nD
esir
eF
olk
854d
47e
870c8
c3f6
4f5
795d
5d
14b
c53c
Sp
anis
hH
arle
mIn
ci..
.B
obD
yla
nA
noth
erS
ide
of
B..
.F
olk
1a2ef
a277ad
42a5ef
ba1794f7
f290ce
2
Sp
rin
gIs
Her
eB
arn
eyK
esse
lT
he
Poll
Win
ner
s..
.Jazz
2c7
46ee
5248a88d
0cf
e244ee
bff
dd
8b
4
To
Be
Alo
ne
Wit
hY
ouB
obD
yla
nN
ash
vil
leS
kyli
ne
Folk
bd
7292a3d
18acb
6ab
7518e1
f47f6
6986
Vis
ion
sof
Joh
ann
aB
obD
yla
nB
lon
de
on
Blo
nd
eF
olk
3d
2c9
0c22a37a3b
b30c2
d9fd
1d
fb988e
Wh
o’s
Th
atC
reep
in’?
Big
Bad
Vood
oo
Dad
dy
Th
isB
eau
tifu
lL
ife
Sw
ing
53122352d
0e6
d9ec
bae9
435fe
ef218ce
Wit
hG
od
On
Ou
rS
ide
Bob
Dyla
nT
he
Tim
esT
hey
Ar.
..F
olk
7740d
47427258176a750d
edaf4
917ed
7
You
’re
AB
igG
irl
Now
Bob
Dyla
nB
lood
On
Th
eT
rack
sF
olk
db
b22e
1e5
3c3
d0c3
fb079158d
318ce
53
IW
ill
Pla
yM
yG
ame.
..B
ran
dN
ewD
eja
Ente
ndu
Pu
nk
16cb
b6959e8
7b
210aaae0
dc1
4cb
372a5
92A
vail
4A
MF
rid
ayP
un
k32e6
d44
9b
6f3
404353ab
bf0
fe40ea
8ef
72
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Dre
amer
Atm
osp
her
eW
hen
Lif
eG
ives
Y..
.H
ip-H
op
ab
4ad
954acc
06d
d92ec
564804f5
15b
40
Gu
ns
and
Cig
aret
tes
Atm
osp
her
eL
ucy
Ford
,T
he
At.
..H
ip-H
op
1105732d
cd789988a5117a566d
40ce
12
Blu
eB
eard
Ban
dO
fH
orse
sIn
fin
ite
Arm
sIn
die
8253fc
b5c6
c8afa
658d
795ce
2aa3f9
51
Bea
uti
ful
Atm
osp
her
eS
ad
Clo
wn
Bad
Win
...
Hip
-Hop
de1
7776
4d
819841485d
7a267a6d
76fd
b
Bre
akY
our
Fra
me
Bad
Ast
ron
au
tH
ou
ston
:W
eH
ave
...
Pu
nk
165ca
85e5
23b
6d
99d
1c0
dcf
cd1c6
b5e9
Bro
ken
Bad
Rel
igio
nT
he
Pro
cess
Of
Be.
..P
un
k5a49305d
c77593b
6298639f7
1c7
466b
1
Com
pli
cati
ons
Atm
osp
her
eO
verc
ast
!H
ip-H
op
60f8
670d
06ce
9079fa
1301e1
1ab
138d
3
Det
lef
Sch
rem
pf
Ban
dof
Hor
ses
Cea
seto
Beg
inIn
die
64a0b
2a2d
bc7
cfc4
64e9
619858d
09ae6
Evil
Sta
teA
ud
iocr
ush
So
You
Call
Th
ese.
..In
die
e28e8
c09628731d
e4e6
ab
9156c0
e6580
Fal
lA
par
tA
vail
Ove
rT
he
Jam
esP
un
ke7
afe
36e0
cdaa8c7
9b
121b
9871e0
740a
Fal
lIn
AR
iver
Bad
lyD
raw
nB
oyT
he
Hou
rO
fB
ewil
...
Ind
ie3b
b776c0
063a12fa
cba36d
521f7
f6d
af
AF
ighti
ng
Wor
dA
sT
all
As
Lio
ns
Blo
od
An
dA
ph
ori
sms
Ind
ieec
5f5
739504cc
b7338960d
90ee
6a1604
F**
*Y
ouL
ucy
Atm
osp
her
eG
od
Lov
esU
gly
Hip
-Hop
b38ce
4f338523fe
f01d
38694d
1acc
a62
Hap
pym
ess
Atm
osp
her
eS
ad
Clo
wn
Bad
Sp
r...
Hip
-Hop
d6315f5
96e3
5a35877211d
a2af7
d675e
Hea
rtA
Tac
tK
idD
yn
amit
e88
Fin
ger
sL
ou
ie..
.P
un
k02d
0ae6
3c0
9b
4b
43ee
4d
27ec
201b
bb
e8
Hel
itro
pe
At
Th
eD
rive
-In
VA
YA
Pu
nk
62455491d
75296f1
9299962331692590
Lev
eled
Ava
ilO
ne
Wre
nch
Pu
nk
da9cf
e1c3
e6c3
20894d
48b
80ab
5c0
c10
Liv
ing
Eac
hD
ayL
ik..
.A
trey
uS
uic
ide
Note
san
d..
.M
etal
f6f0
0856f7
3d
c5e5
5b
295ca
f004d
2826
Lov
e,L
ove,
Lov
e(L
...
As
Tal
lA
sL
ion
sA
sT
all
As
Lio
ns
Ind
ieb
cd4ad
e91ed
c3552273ed
dee
5b
c002c7
Moos
hB
igw
igA
nIn
vit
ati
on
To
...
Pu
nk
1835aeb
2af8
356092522508a25b
c080a
73
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Refl
ecti
ons
Atm
osp
her
eS
even
’sT
ravel
sH
ip-H
op
a5159d
f935ae3
43182161e2
49aaea
f33
Rel
ease
Th
eH
oun
ds
Ban
eG
ive
Blo
od
Pu
nk
eb763a1
bd
2b
2d
5b
1d
3811ff
b545028c9
Sle
epw
alk
Cap
sule
sA
tT
he
Dri
ve-I
nR
elati
on
ship
Of
C..
.P
un
k47f3
cb1034d
e8d
d003266d
e56a49ea
7d
Son
gA
vail
Dix
ieP
un
k1d
10965
f3939d
b1787fb
52f1
b58b
338f
Su
per
her
oB
ane
Hold
ing
Th
isM
om
ent
Pu
nk
25f9
1664ac7
0462eb
edca
5ef
60d
4f7
02
T.V
.A
uto
mat
ic7
Au
tom
ati
c7
Pu
nk
ba1254f
9b
ca069b
091e8
067b
059d
bd
e4
Th
eF
un
eral
Ban
dO
fH
orse
sE
very
thin
gA
llT
h..
.In
die
cf9501a0ad
9e1
383d
e5b
287855035413
Th
eP
aint
Ch
ips
Aw
ayB
ane
ItA
llC
om
esD
own
...
Pu
nk
6ef
72485a718363934b
e3b
b62a1b
3f7
8
Th
eyL
ook
edL
ike
St.
..B
aysi
de
Bay
sid
eP
un
k21a068d
56346acc
90a9d
b506b
5a3a63c
Tin
yV
oice
sB
adR
elig
ion
Str
an
ger
Th
an
Fic
...
Pu
nk
f3744d
82b
0f6
5ee
714a6a749999ea
067
Tom
orro
wB
adR
elig
ion
Gen
erato
rP
un
k58c0
1b
10c0
ace
dc9
988277f5
ad
7a043b
Unti
lW
eF
all
Au
dio
slav
eR
evel
ati
on
sA
lter
nati
veb
8e3
32d
f816a562a226a94f3
c44c8
40b
Wh
enB
adR
elig
ion
Su
ffer
Pu
nk
e212c1
9e1
6b
636b
3530cc
b8506194f5
6
Mat
tres
sA
tmos
ph
ere
Sad
Clo
wn
Bad
Su
m..
.H
ip-H
op
7824cd
0313681fb
b9f7
e1850053b
73b
f
Mu
sica
lC
hai
rsA
tmos
ph
ere
You
Can
’tIm
agin
e...
Hip
-Hop
064ce
062b
471782a15174e5
285ec
944e
Ju
stF
orS
how
Atm
osp
her
eT
he
Fam
ily
Sig
nH
ip-H
op
d30e4
62afa
6e8
6a0d
1078c9
650ce
f2ce
Blu
esfo
ra
Pla
yb
oyB
arn
eyK
esse
lM
usi
cT
oL
iste
nT
oJazz
4ef
2fd
901f4
7585804acf
67389b
6b
897
Eas
yL
ivin
gB
arn
eyK
esse
lP
oll
Win
ner
sT
hre
eJazz
25d
3702
32ef
e012cc
39b
63d
207258d
11
Go
Eas
y(S
eeth
eL
ove)
As
Tal
lA
sL
ion
sY
ou
Can
’tT
ake
It..
.In
die
910634fe
1260a6b
8c2
42710f5
e89f2
09
God
Dec
idin
gH
otW
ater
Mu
sic
Alk
ali
ne
Tri
o,
Ho..
.P
un
k1c9
46f5
49405993700a509982494e8
ed
74
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
How
Lon
gH
asT
his
B..
.B
arn
eyK
esse
lK
esse
lP
lays
Sta
n..
.Jazz
bee
971c
2290d
28c1
0ee
d91e9
3e8
90d
64
Mak
eS
omeo
ne
Hap
py
Bar
ney
Kes
sel
Poor
Bu
tter
fly
Jazz
dc9
75e9
e7f6
893842e2
f0a91a085d
f67
Tig
erR
agB
arn
eyK
esse
lL
et’s
Cook!
Jazz
8b
59781
0b
e5d
182429037cb
3a1b
45f1
e
Th
eB
est
Day
Atm
osp
her
eT
oA
llM
yF
rien
ds.
..H
ip-H
op
0c4
cc79112ed
3ae2
e02f5
854247d
eee1
AS
kel
eton
InT
he
C..
.A
nth
rax
Am
on
gT
he
Liv
ing
Met
al
04ad
80b
d5d
4c6
052e3
73ff
29693d
7c8
1
Acr
obat
As
Tal
lA
sL
ion
sL
afc
ad
ioIn
die
421ec
935f4
32b
1611a28b
3b
4f4
684fd
0
All
On
Bla
ckA
lkal
ine
Tri
oG
ood
Mou
rnin
gP
un
kcc
7d
7f4
4415297b
10aa5a49ee
df7
f0a8
An
oth
erIn
noce
nt
Gir
lA
lkal
ine
Tri
oF
rom
Her
eto
Infi
...
Pu
nk
cae0
ccd
cba8357c7
8745b
9fe
8b
3484b
d
As
You
Wer
eA
lkal
ine
Tri
oG
od
dam
nit
Pu
nk
68f5
a6eb
3a1d
a4d
a28ac4
263604d
6a16
Det
hb
edA
lkal
ine
Tri
oC
rim
son
Pu
nk
f8150e3
0180132ab
e4b
65eff
16a8c9
6b
Dow
nIn
AH
ole
Ali
ceIn
Ch
ain
sM
TV
Un
plu
gged
Alt
ern
ati
ve5ac8
81b
e5d
4912330716b
8f8
af9
61289
Fix
Ava
ilL
ive
at
the
Bott
o..
.P
un
k3f0
e62c7
79c0
8c8
aa5d
9d
fe712ac8
4d
d
Hea
dC
reep
sA
lice
InC
hai
ns
Ali
ceIn
Ch
ain
sA
lter
nati
ve5523767a343ab
9b
54c1
4e1
fc3f5
f3e9
1
IC
an’t
Rem
emb
erA
lice
InC
hai
ns
Face
lift
Alt
ern
ati
veb
7f6
728
e4ee
3d
8956d
9d
26258a1410b
1
My
Fri
end
Pet
erA
lkal
ine
Tri
oA
lkali
ne
Tri
oP
un
kc8
996ef
44e1
e8fb
fec6
259b
5ac0
5198f
Red
eye
All
Els
eF
ails
Th
eC
hase
Of
Gain
Pu
nk
674b
1fb
c821b
c09222ca
3b
343a7d
ac1
e
Sic
km
anA
lice
InC
hai
ns
Dir
tA
lter
nati
veb
d241aa1cf
34cb
dd
a16f5
8963694761e
Sle
epyh
ead
Alk
alin
eT
rio
May
be
I’ll
Catc
h..
.P
un
k63b
40ab
efff6
8c3
1837105395afc
d5b
3
Th
eA
mer
ican
Scr
eam
Alk
alin
eT
rio
Th
isA
dd
icti
on
Pu
nk
a2f4
6c3
2f7
f5d
374252a4f9
c5ee
8933a
Was
Th
ere?
All
anH
old
swort
hR
oad
Gam
esJazz
8d
670e2
0333c5
dcc
3b
161148ea
733acc
75
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Wh
ale
&W
asp
Ali
ceIn
Ch
ain
sJar
Of
Fli
esA
lter
nati
vea4a4022ca
1449ea
9322a55fc
375af3
3a
Wh
oC
ares
Win
sA
nth
rax
Sta
teO
fE
up
hori
aM
etal
9774b
942066508e8
80ab
2fc
e2c3
29306
Zon
eA
llan
Hol
dsw
ort
hA
llN
ight
Wro
ng
Jazz
35cd
7f2
95150f3
9110b
0b
134ed
a627b
f
Cla
vic
leA
lkal
ine
Tri
oD
am
nes
iaP
un
kfc
466622599b
562b
220f3
c8482c6
00f7
IF
oun
dA
way
Alk
alin
eT
rio
Agony
An
dIr
ony
Pu
nk
d6af0
9cd
7e0
a87835f1
f8c3
5ab
ce112f
Bor
ne
onth
eF
MW
av..
.A
gain
stM
e!N
ewW
ave
Pu
nk
97c0
f077ed
ab
a7eb
1a3a6555d
f2d
a5c3
Esp
irit
uA
lD
iM
eola
,P
aco
D..
.T
he
Gu
itar
Tri
oJazz
f7e7
6c8
e9aca
49e0
fb1e1
5005aac5
7c1
Gre
edy
All
Mass
Ner
der
Pu
nk
358a8f7
1c9
af2
cb4b
17f8
77f6
116a705
ItR
eall
yS
uck
sW
he.
..A
gain
stA
llA
uth
ori
tyD
estr
oyW
hat
Des
t...
Pu
nk
cbff
7407
7b
3851cf
20f6
f5b
3559656c0
Kev
inS
econ
ds
-E
xt.
..K
evin
Sec
ond
sS
pli
tP
un
k4894af2
1c7
7456964a0d
4602d
5640d
68
Ric
ean
dB
read
Aga
inst
Me!
As
Th
eE
tern
al
Co..
.P
un
k4b
55f1
2772e7
6f4
875f0
fcb
113d
58ac3
Sco
rch
Ale
xS
kol
nic
kT
rio
Tra
nsf
orm
ati
on
Jazz
9f1
0c7
2ce
52b
6e3
0429827f7
cbaf7
e9a
Stu
nn
ing
Up
onA
gain
stT
omorr
ow’s
Sky
Ju
mp
Th
eH
edges
F..
.In
die
e00d
22f
3f2
f93e6
7a7e2
2e3
62424972d
Th
eC
olle
geT
ryA
gain
stT
omorr
ow’s
Sky
Th
eL
ost
Tap
esIn
die
a49d
c3b
be6
36f0
fff0
31a267657e9
082
Jor
dan
’sF
irst
Ch
oice
Aga
inst
Me!
Rei
nve
nti
ng
Axl
Rose
Pu
nk
80b
a058
f53431f5
5d
7b
738b
e7d
f93249
Vio
len
ceA
gain
stM
e!S
earc
hin
gfo
ra
F..
.P
un
k547879aaa25494228ad
733d
69d
22eb
a0
Cer
eal
War
sA
FI
An
swer
Th
at
An
dS
...
Pu
nk
ea611937f3
6f1
6e9
bd
40762769fd
0419
Dea
thO
fS
easo
ns
AF
IS
ing
Th
eS
orr
owP
un
k56804d
635133aa0cd
61b
0fa
5a9a45fe
0
Ret
irin
gA
Wil
hel
mS
crea
mM
ute
Pri
nt
Pu
nk
8b
de5
e7c2
074fb
78021f0
ea728863929
Tob
yA
gain
stA
llA
uth
ori
tyA
llF
all
Dow
nP
un
k5fb
54c8
9b
18d
b6655c7
ed40d
165f6
c4d
76
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Bli
nk
88F
inge
rsL
ou
ieB
ehin
dB
ars
Pu
nk
0b
4459d
06085b
d900fd
122e7
996619d
a
Con
grat
ula
tion
sA
Wil
hel
mS
crea
mR
uin
erP
un
k18d
2f2
9d
b3f8
09f8
317ae2
d7d
05a41d
a
No
Poet
icD
evic
eA
FI
Bla
ckS
ail
sIn
Th
...
Pu
nk
22c9
b5a
f6c3
a741d
1f2
bd
ed37ad
14b
39
Th
isS
ecre
tN
inja
AF
IV
ery
Pro
ud
Of
Ya
Pu
nk
ca8138537b
ca4813fd
844d
143a7eb
6f2
Th
omas
AP
erfe
ctC
ircl
eM
erd
eN
om
sM
etal
93ce
1a6a4445ab
a32d
afe
e7f8
880d
d08
Ou
rG
hos
ts(c
onte
mp
...
AW
ilh
elm
Scr
eam
Care
erS
uic
ide
Pu
nk
1b
e791b
5928ad
47525707b
eddc9
88f5
e
F**
*T
his
Pla
ceF
righ
ten
edR
ab
bit
AF
righte
ned
Rab
b..
.In
die
a695f4
6f9
46e0
a155aaa7067b
d499f8
9
Aft
erT
he
Lau
ghte
r..
.An
dY
ouW
ill
Kn
o..
.S
ou
rce
Tags
&C
od
esIn
die
67004931ad
c13036cd
a4c2
fc3467e0
3a
Col
ourb
lin
d7
Sec
ond
sT
he
Cre
wP
un
k6fd
7189
f514b
a5a07ed
135ff
cab
197e9
Tak
eit
and
run
Dro
pkic
kM
urp
hys
Th
eW
arr
ior’
sC
od
eP
un
k74a5b
819555ea
cb9878b
38b
13a21af1
6
Bas
tard
sO
nP
arad
eD
rop
kic
kM
urp
hys
Bla
ckout
Pu
nk
b4210b
41b
82634d
61f1
1c1
a7f9
7261b
5
Sta
ind
Gla
ssA
nd
Ma.
..R
ise
Aga
inst
Un
rave
lin
gP
un
kca
08ce
5c3
559cd
1c4
a96b
45e4
fe92b
b5
Hig
hP
ress
ure
Low
Aga
inst
Me!
Wh
ite
Cro
sses
Pu
nk
a725d
cb1d
d072e8
f05d
8b
2215cc
41565
Ch
eek
To
Ch
eek
Geo
rge
Van
Ep
sL
egen
ds:
Solo
Gu
i...
Jazz
5d
59ad
03488f7
562528e1
222b
d758d
c9
Lie
toM
eD
epec
he
Mod
eS
om
eG
reat
Rew
ard
New
Wav
e6eb
0a73
a43d
8b
f14f9
91436421011633
Peo
ple
are
Peo
ple
(...
Dep
ech
eM
od
eS
om
eG
reat
Rew
ard
New
Wav
ee0
a60609fa
1838f0
95831b
296078ea
4c
See
You
(Liv
ein
Ha.
..D
epec
he
Mod
eA
Bro
ken
Fra
me
New
Wav
e6762e4
3e4
6f9
33d
a1e3
bd
d027d
1586f7
Sh
out
Dep
ech
eM
od
eS
pea
k&
Sp
ell
New
Wav
ec1
88f9
f1091d
99326264624074728fa
8
Wor
kH
ard
Dep
ech
eM
od
eC
on
stru
ctio
nT
ime.
..N
ewW
ave
a37b
7b
165911b
a2e1
883154327542fd
5
Cra
nk
(Liv
e)C
ath
erin
eW
hee
lW
ishvil
le(B
onu
s..
.A
lter
nati
veb
64a881
917298607e7
652c6
63d
7cc
667
77
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Wil
lin
gT
oW
ait
Cat
her
ine
Wh
eel
Ma
Soli
tud
a[D
isc
2]
Alt
ern
ati
vecff
0f8
a757cd
3190686c2
7cb
3cb
63aa7
Th
eR
oos
ter
Atm
osp
her
eS
ad
Clo
wn
Bad
Fal.
..H
ip-H
op
75b
1a29
de7
6e3
006d
c9b
9e2
3f1
7d
117a
All
IsF
ull
Of
Lov
eD
eath
Cab
for
Cu
tie
Th
eS
tab
ilit
yE
PIn
die
9c0
0d
7bb
aeb
662ae4
6a4695588e3
818f
Th
eT
hin
gsT
hat
Hat
...
Atm
osp
her
eS
tric
tly
Lea
kage
Hip
-Hop
e7ce
34aa696b
6ab
1d
17ce
7780e5
96417
Wh
ite
Noi
seA
tmos
ph
ere
Lea
kA
tW
ill
Hip
-Hop
8a6205b
97ee
03858a2b
ff46323d
ef534
Su
dd
enD
eath
InC
ar..
.B
ran
dN
ewY
ou
rF
avori
teW
ea..
.P
un
k4f8
eaca
782d
d7f1
0c5
b4e2
e246910b
29
Car
efu
lN
owC
opel
and
Eat,
Sle
ep,
Rep
eat
Ind
ie43d
4aa5
fc4b
82605161b
c9a9c7
0c8
10d
Sou
lM
eets
Bod
yD
eath
Cab
for
Cu
tie
Pla
ns
Ind
ie8ab
447d
2d
2894aeb
fd9ee
6e6
c7438b
1f
Sw
eete
stP
erfe
ctio
nD
epec
he
Mod
eV
iola
tor
New
Wav
e046f3
e9b
827c1
229c2
cb9633b
29f9
041
Nev
erB
ough
tIt
Din
osau
rJr
Han
dIt
Ove
rA
lter
nati
ve8f6
7d
7e6f4
672c5
0f1
b32d
e6c4
014ca
2
Fee
lT
he
Pai
nD
inos
aur
Jr
Wit
hou
ta
Soun
dA
lter
nati
ve0e8
b9d
f549afc
2c2
e221583459d
9776b
Sta
rtC
hop
pin
Din
osau
rJr.
Wh
ere
You
Bee
n?
Alt
ern
ati
vec9
b2972
a74e7
393c5
9554ce
43109d
8cb
Day
Tw
oE
xp
losi
ons
InT
he
Sky
Tra
vels
InC
on
sta..
.In
die
fe219b
7526409c4
bd
91630d
21014ad
75
You
An
dT
he
Nig
ht
A..
.Jam
ieC
ull
um
Poin
tles
sN
ost
alg
icJazz
44b
2b
4e1
00ea
c6e5
f2502e0
f65624fd
9
For
Jac
oJoh
nM
cLau
gh
lin
Ind
ust
rial
Zen
Jazz
eceb
075
22e9
2b
152fd
10a35b
9d
2692aa
Aft
erth
eF
act
Joh
nS
cofi
eld
Qu
iet
Jazz
d0ea
038
32f8
e3d
24343e3
c6222316d
13
(Beg
inn
ing
ofW
hat
’...
Kyu
ssW
retc
hS
ton
erM
etal
606689b
c8b
e37893d
5f0
58d
42322f0
af
100/
Sp
ace
Cad
et/D
em..
.K
yu
ssW
elco
me
toS
ky
Va..
.S
ton
erM
etal
981ec
afc
9fc
468804b
c5e5
5ea
e41e4
f5
50M
illi
onY
ear
Tri
...
Kyu
ssB
lues
for
the
Red
...
Sto
ner
Met
al
1601d
86145ee
de4
c5e5
e959d
0368b
456
AD
ayE
arly
and
aD
...
Kyu
ssM
uch
as
Gra
cias:
T..
.S
ton
erM
etal
cb944f8
0a2d
a9a24b
6b
290868464b
be2
78
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Bor
nto
Hu
laK
yu
ssQ
uee
ns
of
the
Sto
...
Sto
ner
Met
al
ca7fb
5bfd
2d
f8e2
dd
e7d
62d
39c4
d816b
Cat
amar
anK
yu
ss..
.An
dth
eC
ircu
s...
Sto
ner
Met
al
45f2
07ff
972ad
30ed
272468c4
d84a78b
Ab
ove
the
Tre
etop
sP
atM
eth
eny
Sec
ret
Sto
ryJazz
a284b
9a8ca
c86b
ce4ec
2b
9995048d
9d
d
Bra
inof
J.
Pea
rlJam
Yie
ldA
lter
nati
ve7977b
de2
149e5
1a70cd
a4b
aafc
4b
f318
Hey
You
Pin
kF
loyd
Th
eW
all
Rock
6ae0
8936b
bd
9695e9
95e7
1744b
290d
8a
InT
he
Fle
shP
ink
Flo
yd
Th
eW
all
Rock
014470cc
b0e6
f192d
1e5
44e4
0d
c708c5
1492
Cou
nti
ng
Cro
ws
Satu
rday
Nig
hts
&..
.A
lter
nati
vefc
eac5
6556090c5
96a1361d
8d
d97c3
db
Fol
low
Th
eR
eap
erC
hil
dre
nof
Bod
om
Foll
owT
he
Rea
per
...
Met
al
16ef
5a10094e4
81045f7
1c1
de1
c149ab
10.4
5A
mst
erd
amC
on..
.F
un
eral
For
AF
rien
dS
even
Way
sT
oS
cr..
.P
un
kd
bfa
97ff
88b
0a2374c8
e533ae6
32d
87e
45E
lvis
Cos
tell
oW
hen
IW
as
Cru
elR
ock
1659d
33909d
5ce
658b
ef5985f8
0d
9a27
6-00
Dre
amT
hea
ter
Aw
ake
Pro
g.
Met
al
07a74e0
3b
e2ef
8cf
acc
fdf5
f4e6
d44d
7
AP
ain
that
I’m
Use
...
Dep
ech
eM
od
eP
layin
gth
eA
ngel
New
Wav
e360b
59c
d45981c0
84959b
8488a0a4a2b
All
My
Lif
eF
oo
Fig
hte
rsO
ne
By
On
eA
lter
nati
ve92f1
3e3
a70a50324aab
c5488d
4856ed
a
Ap
oca
lyp
seN
owan
d..
.E
very
Tim
eI
Die
Gu
tter
Ph
enom
enon
Met
al
7d
b4eb
b50d
91ae9
299c4
c54b
dcc
f9f7
2
Arr
ows
Fir
ewor
ks
Gosp
elP
un
k0f4
ba52
3f3
eefc
aca
d1f1
701576e1
e67
Be
Cal
mF
un
.A
iman
dIg
nit
eIn
die
1641aa42c7
4ac8
c18fe
a983301369727
Ben
dto
Squ
ares
Dea
thC
abfo
rC
uti
eS
om
eth
ing
Ab
ou
tA
...
Ind
ied
472b
8b
81d
0702b
6fb
6ab
e76e2
90d
639
Bey
ond
Bel
ief
Elv
isC
oste
llo
Imp
eria
lB
edro
om
Rock
190ad
3131987729aa260ad
1d
8afd
d140
Bil
lO
fG
ood
sF
ace
To
Fac
eH
owT
oR
uin
Eve
ry..
.P
un
k17611d
31b
cf93e3
f8ca
a29f8
0364e6
cc
Blo
md
ido
Diz
zyG
ille
spie
Bir
k’s
Work
sJazz
c1d
d7d
a7ed
8e3
708085f7
8d
401211698
79
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Cat
apu
ltC
ounti
ng
Cro
ws
Rec
over
ing
Th
eS
a..
.A
lter
nati
vecf
6261c6
ac1
3d
e52d
dfa
7d
acf
eaf0
613
Tu
rnC
hri
stie
Fro
nt
Dri
ve
Anth
olo
gy
Ind
ie39111d
ea88c8
295ad
a614f5
37f1
878ee
Col
dA
sIc
eF
orei
gner
Th
eD
efin
itiv
eC
o..
.R
ock
7ff
0d
82b
8f7
01ee
05aca
bafd
c5e9
73b
e
Bri
ghte
stC
opel
and
Ben
eath
Med
icin
e..
.In
die
ab
73b
0ca
526452ce
950fd
1c6
229fc
347
Daa
hou
nd
Em
ily
Rem
ler
East
To
Wes
Jazz
d76199f
8f5
41b
40639a52902aac9
4e4
1
Bar
rel
ofa
Gu
nD
epec
he
Mod
eU
ltra
New
Wav
e4609d
255a061cd
a985482865282b
326b
Bla
ckC
eleb
rati
onD
epec
he
Mod
eB
lack
Cel
ebra
tion
New
Wav
ecf
c28ac9
8d
0b
64f3
7812a78ee
31d
8746
Dre
amO
nD
epec
he
Mod
eE
xci
ter
New
Wav
ed
68a98b
bf2
eb70b
1118a706d
e2a32960
IF
eel
You
Dep
ech
eM
od
eS
on
gs
of
Fait
han
...
New
Wav
e6f5
3ce
05b
702b
b615fb
c6a0ca
79ef
eb4
Dia
mon
dL
ight
Boog
ieC
her
ryP
opp
in’
Dad
die
sS
ou
ldC
ad
dy
Sw
ing
9a43165d
2a0d
ce52f6
ee53c1
f23f4
f5e
For
get
the
Sw
anD
inos
aur
Jr.
Din
osa
ur
Alt
ern
ati
vee1
db
c5c8
3aafc
7182a8969d
4d
6725c7
0
Fre
akS
cen
eD
inos
aur
Jr.
Bu
gA
lter
nati
vea6f9
cfc4
5b
4cd
a1423e2
6587d
78406d
0
ID
on’t
Th
ink
So
Din
osau
rJr.
ID
on
’tT
hin
kS
oA
lter
nati
ved
9c9
5f3
ef8d
1f4
3aed
19b
f25f6
aa6d
f3
Lit
tle
Fu
ryT
hin
gsD
inos
aur
Jr.
Foss
ils
Alt
ern
ati
vec1
47fa
8766494aa765508c6
78f7
a732e
Wh
atev
er’s
Cool
Wit
...
Din
osau
rJr.
Wh
ate
ver’
sC
ool
W..
.A
lter
nati
ved
56497e
19a237cf
35211fc
0980544a45
Dol
lars
An
dS
ense
Fad
edG
rey
AQ
uie
tT
ime
Of
D..
.P
un
k58e6
137d
65d
5a1fa
b96f7
70e9
a3cf
4a0
Fam
ous
For
Not
hin
gD
rop
kic
kM
urp
hys
Th
eM
ean
est
Of
Ti.
..P
un
k60b
0492
f3d
5f9
f05fa
5b
28c6
0821ed
a9
Dru
nk
Dad
dy
Ch
erry
Pop
pin
’D
ad
die
sF
eroci
ousl
yS
ton
edS
win
g20d
4237
4955509b
0d
f8575648855d
de9
Fir
stB
reat
hA
fter
...
Exp
losi
ons
InT
he
Sky
Th
eE
art
hIs
Not
...
Ind
ie8f2
4b
7343ab
be5
0809310d
7d
3554ec
71
Th
eB
irth
and
the
D..
.E
xp
losi
ons
InT
he
Sky
All
of
aS
ud
den
I...
Ind
ie9191b
37e3
570c4
6fc
115986264781570
80
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
You
’ve
Don
eN
othin
gF
ace
To
Fac
eD
on
’tT
urn
Aw
ayP
un
ke8
26f8
49237e4
cb4a5951c8
e9ef
7c0
2c
Fam
ily
Syst
emC
hev
elle
Won
der
Wh
at’
sN
ext
Met
al
5508ec
d353ab
ffacc
dc6
10d
c97cc
f686
Fro
mF
inis
hto
Sta
r...
Cou
nte
rfit
Fro
mF
inis
hto
St.
..In
die
d68d
290fb
76d
f0c9
408b
51d
e63b
cc655
Fro
mW
est
Tex
asE
xp
losi
ons
InT
he
Sky
Fri
day
Nig
ht
Lig
hts
Ind
iea5513ca
c22d
a1712d
41942c0
67b
a5572
Th
eM
oon
IsD
own
Fu
rth
erS
eem
sF
ore
ver
Hop
eT
his
Fin
ds
Y..
.In
die
a62b
faf4
df9
bad
9a4810ace
93373111f
Good
Bai
tD
izzy
Gil
lesp
ieD
izzy
Atm
osp
her
eJazz
33c9
a8b
a09e3
782835fa
e910ed
127d
3e
Gre
etD
eath
Exp
losi
ons
InT
he
Sky
Th
ose
Wh
oT
ell
Th
...
Ind
ieb
61b
7b
2fe
de5
78b
05b
b823d
a36309d
94
Gro
ovin
’H
igh
Ch
arli
eP
arker
Yard
bir
dS
uit
eD
i...
Jazz
b2a3c4
3ec
25314fd
6f7
bf7
5e3
bb
19807
Han
gin
arou
nd
Cou
nti
ng
Cro
ws
Th
isD
eser
tL
ife
Alt
ern
ati
veb
1549e4
db
b6f6
f1a24b
24f8
7a326ff
7a
Har
dC
and
yC
ounti
ng
Cro
ws
Hard
Can
dy
Alt
ern
ati
vee5
d812b
346c9
ac2
4a729075d
a9d
e9ce
b
Hea
ven’s
Not
Ove
rfl..
.C
orro
sion
Of
Con
for.
..D
eliv
eran
ceS
ton
erM
etal
4b
7b
1152b
53057d
54fd
0c2
55d
ad
01cc
1
Hol
din
gS
omeo
ne’
sH
...
Cir
caS
urv
ive
Ju
turn
aIn
die
9b
71f8
8886664f3
faff
882b
83ac5
c034
Hom
eIs
aF
ire
Dea
thC
abfo
rC
uti
eC
od
esan
dK
eys
Ind
ie30f5
f4d
a9a3c6
4b
dab
16d
b103e9
22f9
b
How
To
Sta
rtA
Fir
eF
urt
her
See
ms
Fore
ver
How
To
Sta
rtA
Fir
eIn
die
fa90934a29eb
6c5
3424601339d
14b
e49
IF
eel
Fre
eC
ream
Th
eB
est
Of
Cre
am
...
Rock
c06e0
2ef
0ae9
cb56a18e2
8f7
2d
30c9
c8
IfT
hey
Mov
e...
Kil
l...
Fai
rwea
ther
IfT
hey
Mov
e...
Ki.
..P
un
kef
0379d
31ed
555a440c3
522c4
2f9
d572
Imm
ure
Dam
iera
MU
SIC
Ind
iee1
343d
447d
98cc
53b
2eb
58aec
bfa
cf0a
Ina
Sen
tim
enta
lM
ood
Du
keE
llin
gton
Bes
tof
Du
keE
lli.
..Jazz
e7d
1f3
e241250ef
4b
35158e7
6a1c0
b98
InC
hai
ns
Dep
ech
eM
od
eS
ou
nd
sO
fT
he
Un
i...
New
Wav
e56e0
cf883a20203f7
1c3
5cd
cacf
4b
9c8
Invo
cati
onO
fA
poca
...
Dra
gon
For
ceV
all
eyO
fT
he
Dam
ned
Met
al
6d
6a3d
7315f7
dd
2b
455c8
ea53c9
6ef
5b
81
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Thu
mb
JM
asci
sM
art
in+
Me
Alt
ern
ati
ve2d
d6c6
456064d
49e2
db
90cf
5ce
d01d
67
Kee
pin
gth
eB
lad
eC
ohee
dan
dC
am
bri
aG
ood
Ap
oll
o,
I’m
...
Pu
nk
e5ea
ec7e3
33e8
9c3
20902b
92fb
a3a165
Kid
sO
nT
he
Str
eet
Ch
erry
Pop
pin
’D
ad
die
sK
ids
on
the
Str
eet
Sw
ing
ae4
a7b
c4f4
971360246d
ad
3ae0
01fe
1b
Kin
gof
the
Rot
ten
Cor
rosi
onO
fC
on
for.
..W
iseb
lood
Sto
ner
Met
al
522acf
2e8
dc0
fc472b
f694d
ac3
e317b
b
Kis
sO
fD
eath
Dok
ken
Back
For
Th
eA
ttack
Met
al
30aee
34b
6e5
1158a917ce
d251195d
9f7
Lig
ht
Up
Ah
ead
Fu
rth
erS
eem
sF
ore
ver
Hid
eN
oth
ing
Ind
ied
5f1
b746e3
43b
715f5
0989083c2
6681a
Liv
ing
Dea
dB
eat
Ch
ild
ren
ofB
od
om
Are
You
Dea
dY
et?
Met
al
9b
cbb
0762fd
ab
349e3
1aff
3213e7
f049
My
Sp
irit
Wil
lG
oO
nD
rago
nF
orce
Son
icF
ires
torm
Met
al
bd
9399f4
063278b
2025e6
f6b
73ae7
bc5
Nee
dle
d24
,7C
hil
dre
nof
Bod
om
Hate
Cre
wD
eath
roll
Met
al
06b
5ef
31d
23d
148682c4
1b
41d
0e4
4d
99
No
Act
ion
Elv
isC
oste
llo
Th
isY
ears
Mod
elR
ock
9a178e5
7651f8
a1577f9
f8d
25a1ad
95d
No
On
eR
eall
yW
ins
Cop
elan
dIn
Moti
on
Ind
ie2ee
d6b
d4a1643291d
57086a8f4
f7960a
On
eE
yed
ea&
Ab
ilit
ies
Fir
stB
orn
Hip
-Hop
2287cd
13f0
90b
b53f9
9eb
071b
fac4
9e2
Op
enU
pan
dW
ail
Chu
ckR
agan
Los
Fel
izIn
die
01e1
c728ff
7a73a22e1
635a38153ec
2a
Oro
,In
cien
soY
Mir
raD
izzy
Gil
lesp
ieA
fro-C
ub
an
Jazz
M..
.Jazz
68e5
ad
778b
4b
5c1
22b
36e6
9d
4b
825018
Ou
rD
elig
ht
Diz
zyG
ille
spie
Cool
Bre
eze
Jazz
48d
0846
3e4
ff1e2
c71d
a7c5
e547e2
421
Per
did
oD
izzy
Gil
lesp
ieS
alt
Pea
nu
tsJazz
e906b
38e1
36711261e7
a475442b
77b
df
Pol
itik
Col
dp
lay
AR
ush
Of
Blo
od
T..
.In
die
5094d
9afd
e5b
aef
58b
06574f5
e8ca
e81
Pre
lud
eC
hio
dos
All
’sW
ell
Th
at
E..
.P
un
k46f7
76d
66344ff
26066b
18f1
6573e2
83
Pu
llM
eU
nder
Dre
amT
hea
ter
Images
And
Word
sP
rog.
Met
al
72a0d
7f72d
6c1
f66a1b
9e9
1f6
aef
ceb
d
Reg
ress
ion
Dre
amT
hea
ter
Met
rop
oli
sP
t.2
S..
.P
rog.
Met
al
2e4
813801b
fee3
4cb
944606b
e0c2
5ac3
82
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Rom
eoA
Go-
Go
Eve
ryT
ime
ID
ieH
ot
Dam
n!
Met
al
b6413d
577e8
f7d
241a0cc
08b
c096b
e29
Rou
nd
Her
eC
ounti
ng
Cro
ws
Au
gu
stan
dE
very
t...
Alt
ern
ati
ve886898a1389b
47b
fb2b
fb740d
7d
557c7
Sch
ool
dD
ays
Diz
zyG
ille
spie
Imp
rom
ptu
Jazz
a7322b
e4e6
1c7
a274558af9
5e5
3fc
ff7
Scr
app
leF
rom
the
A..
.C
har
lie
Par
ker
Yard
bir
dS
uit
eD
i...
Jazz
f7c0
d45
0774e0
7d
228ce
082f9
f50b
8e8
Sig
ne
Eri
cC
lap
ton
Un
plu
gged
Blu
es454836e1
32d
51901f1
858aab
67ca
f2c8
Slu
nky
(Ori
gin
alm
ix)
Eri
cC
lap
ton
Eri
cC
lap
ton
Del
u..
.B
lues
06cc
086a1640b
454d
e1ad
ae7
deb
41508
Slu
nky
Eri
cC
lap
ton
Eri
cC
lap
ton
Del
u..
.B
lues
bd
80cd
726d
bc0
7f6
fdcb
2e6
34a46c8
10
Som
eN
ights
Intr
oF
un
.S
om
eN
ights
Ind
ie718067f3
ef0fd
f5a51d
565159b
cad
ef9
Sou
ven
irC
ounte
rfit
Su
per
Am
use
men
tM
...
Ind
ied
34fb
07ff
47277e9
46160f5
3684cb
141
Sow
ing
Sea
son
(Yea
h)
Bra
nd
New
Th
eD
evil
An
dG
od
...
Pu
nk
f300251868f7
837d
5ab
b6eb
ffa947502
Ste
adie
rF
oot
ing
Dea
thC
abfo
rC
uti
eT
he
Ph
oto
Alb
um
Ind
ie92f2
f19f4
014c8
d929840b
75ed
9738e9
Sto
ne
Bre
aker
Cor
rosi
onO
fC
on
for.
..In
Th
eA
rms
Of
God
Sto
ner
Met
al
6d
4cd
c6b
183e3
c954f3
63b
186a1fe
18c
Tal
kin
gD
esce
nd
ents
Cool
To
Be
You
Pu
nk
26ac6
c18ee
4b
5e2
40a26e5
228e3
f7150
Th
eB
rill
iant
Dan
ceD
ashb
oard
Con
fess
ion
al
Th
eP
lace
sY
ou
Ha..
.In
die
5128881b
9431f0
92e4
ee4ed
01b
e5828d
Th
eG
reys
Fri
ghte
ned
Rab
bit
Sin
gT
he
Gre
ys
Ind
ie1fb
a10c
e024d
0e2
c5a074eb
0754f2
183
Th
eM
od
ern
Lep
erF
righ
ten
edR
ab
bit
Th
eM
idn
ight
Org
a..
.In
die
077369139c2
1a6af6
81b
097b
423ef
eed
Th
eM
oon
IsD
own
Fu
rth
erS
eem
sF
ore
ver
Th
eM
oon
IsD
own
Ind
ie4a9280f6
d8a01603b
5c8
4e5
d524e9
a60
Th
eN
ewY
ear
Dea
thC
abfo
rC
uti
eT
ran
satl
anti
cism
Ind
ie3a8fb
f6436f9
68f0
35f0
ee90e4
19ae4
a
Th
eR
oot
Of
All
Evil
Dre
amT
hea
ter
Oct
avari
um
Pro
g.
Met
al
df0
a451
a0084f4
809c4
59d
1107c4
c15a
Th
eS
earc
hC
her
ryP
opp
in’
Dad
die
sR
ap
idC
ity
Mu
scle
...
Sw
ing
b55b
5635b
09ae6
439fc
2b
1c6
b4c9
ef3e
83
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Th
ese
Sh
rou
ded
Tem
ple
sC
orro
sion
Of
Con
for.
..B
lin
dS
ton
erM
etal
a7026c1
2eb
3d
0173a55d
b92fa
858ca
63
Th
ings
Fri
ghte
ned
Rab
bit
Th
eW
inte
rof
Mix
...
Ind
ie37465ca
905d
7671e7
6d
31451b
63b
a58d
Tit
leT
rack
Dea
thC
abfo
rC
uti
eW
eH
ave
Th
eF
act
s...
Ind
ied
160106
865773e4
4574d
db
16110d
87b
3
Un
chai
nT
he
Nig
ht
Dok
ken
Un
der
Lock
An
dK
eyM
etal
f23a32d
902263866e2
c65b
38eb
18ff
2a
Wom
enD
efL
epp
ard
Hyst
eria
Met
al
683ef
59f4
eea6c1
fe5f5
1a7883c8
fb12
Zoot
Su
itR
iot
Ch
erry
Pop
pin
’D
ad
die
sZ
oot
Su
itR
iot
Sw
ing
d5733d
e00ca
4e2
9f9
ee229d
1e3
be3
368
AS
ong
For
Ou
rF
ath
ers
Exp
losi
ons
InT
he
Sky
How
Str
an
ge,
Inn
o..
.In
die
8a137d
90ca
daf2
658a56ef
956cc
8a69d
Act
sof
Man
Fri
ghte
ned
Rab
bit
Ped
estr
ian
Ver
seIn
die
5e4
667e8
60999cd
ab
2f6
4aee
789b
69f3
Bad
Boy
frie
nd
Gar
bag
eB
leed
Lik
eM
eA
lter
nati
veb
82ad
fa29b
c7a84a2719a00133ab
48c1
Bix
by
Can
yon
Bri
dge
Dea
thC
abfo
rC
uti
eN
arr
owS
tair
sIn
die
f724938a630a215093a40b
3f8
a5680b
5
Lit
tle
Bri
bes
Dea
thC
abfo
rC
uti
eT
he
Op
enD
oor
EP
Ind
iec2
5a153e2
437d
bb
50fe
3f8
dd
67c0
168e
Nev
erL
etM
eD
own
A..
.D
epec
he
Mod
eM
usi
cF
or
Th
eM
ass
esN
ewW
ave
4e5
4c9
0370b
5b
5015b
8267d
c4223a10f
Rei
ntr
od
uci
ng
Eye
dea
&A
bil
itie
sE
&A
Hip
-Hop
456ab
bf0
b03b
d2375766b
c8e1
1792661
Hay
Fev
erE
yed
ea&
Ab
ilit
ies
By
Th
eT
hro
at
Hip
-Hop
4fc
3d
7afe
bd
9d
bb
a8c2
48665524e9
860
Han
g’E
mH
igh
Dro
pkic
kM
urp
hys
Goin
gO
ut
InS
tyle
Pu
nk
b980c5
15eb
fcb
db
28cc
37fa
050d
f0b
c1
Hel
lhou
nd
sO
nM
yT
rail
Ch
ild
ren
ofB
od
om
Blo
od
dru
nk
Met
al
593afd
cc7a44821d
16c5
7ca
a92a2224d
Las
tK
now
nS
urr
oun
d..
.E
xp
losi
ons
InT
he
Sky
Take
Care
,T
ake
C..
.In
die
72c0
528f9
238b
82e5
365e0
8254806627
Lif
eIn
Tec
hn
icol
orC
old
pla
yV
iva
La
Vid
aIn
die
ae8
a4ef
251c8
f84f3
9013038ee
0144a1
Lov
e,in
Itse
lfD
epec
he
Mod
eC
on
stru
ctio
nT
ime.
..N
ewW
ave
65c6
79ed
e1555a10aaaaed
79034a9fd
1
My
Sec
ret
Gar
den
Dep
ech
eM
od
eA
Bro
ken
Fra
me
New
Wav
e66a436ab
e24e9
d4b
a785e5
983a4f5
df7
84
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
New
Lif
eD
epec
he
Mod
eS
pea
k&
Sp
ell
New
Wav
e95c4
b3b
2b
6eb
7506d
2e3
5076d
65b
92cc
Sat
urd
ayC
hri
stie
Fro
nt
Dri
ve
Ch
rist
ieF
ront
Dri
ve
Ind
ie046b
623
ae6
af0
def
41455ce
be7
d3b
1e5
Sh
ould
Anyth
ing
Go
...
Fac
eT
oF
ace
Lau
gh
Now
...L
au
gh
...
Pu
nk
fa66ff
2a07487b
bcb
8e9
1d
d25e2
82f9
7
Shu
tY
our
Mou
thG
arb
age
bea
uti
fulg
arb
age
Alt
ern
ati
vec1
8486afe
f940563089490d
287c2
9b
0f
Unti
tled
(Lov
eS
ong)
Cou
nti
ng
Cro
ws
Un
der
wate
rS
un
shin
eA
lter
nati
vef0
067d
e21686b
4511d
c66826625ef
3e4
Way
dow
nC
ath
erin
eW
hee
lW
ayd
own
[1]
Alt
ern
ati
ve24cf
7ce
ae0
821c7
16e9
291c5
d0f7
f704
We
Lau
ghIn
door
s(T
...
Dea
thC
abfo
rC
uti
eT
he
Joh
nB
yrd
E.P
.In
die
6c2
3e0
851d
fc46a9d
728d
2b
efb
72b
298
Mylo
Xylo
toC
old
pla
yM
ylo
Xylo
toIn
die
e1110ce
86c0
a5f6
5d
2d
9d
ea2c2
a60670
Sh
ould
You
Ret
urn
Cop
elan
dY
ou
Are
My
Su
nsh
ine
Ind
ied
e0afe
cdc1
dee
a07c4
3649e4
68366d
76
Daa
hou
dE
mil
yR
emle
rR
etro
spec
tive
Vol.
..Jazz
035172b
dca
326d
b82c1
ab
bc1
52c6
e2c5
Ph
otob
oot
hD
eath
Cab
for
Cu
tie
Forb
idd
enL
ove
E.P
.In
die
b69c2
9d6b
9b
a7f0
ec803fa
e7992049c5
Ele
ctra
Mad
eM
eB
lin
dE
verc
lear
Sp
ark
lean
dF
ad
eA
lter
nati
vefd
0844d
82c0
da9d
b567282f7
95732129
Dis
app
ointe
dF
ace
To
Fac
eR
eact
ion
ary
Pu
nk
2f4
9d
6db
f02ace
6873466cb
8a47a241f
Ove
rcom
eF
ace
To
Fac
eIg
nora
nce
isB
liss
Pu
nk
f845b
bf2
9b
e5c1
e321812b
68825a7c1
7
Res
ign
atio
nF
ace
To
Fac
eF
ace
To
Face
Pu
nk
a2084b
725a58fc
da340af3
42b
383e7
c8
Str
ugg
leF
ace
To
Fac
eB
igC
hoic
eP
un
kb
4fa
c71f3
1d
df6
96ad
2aca
43ed
1789b
7
Wal
kT
he
Wal
kF
ace
To
Fac
eL
ive
Pu
nk
66d
07f7
44c0
d743b
3b
77fb
c503a441ad
123
Dro
pF
ace
To
Fac
eT
hre
eC
hord
sA
nd
...
Pu
nk
511c2
eab
7e9
63a111b
0e4
287c2
64a3a0
Geo
grap
hy,
Von
neg
ut.
..F
irew
orks
All
IH
ave
To
Off
...
Pu
nk
31b
2586
4d
b5e1
887c9
96957723ce
b8ae
Hel
ena
Bea
tF
oste
rT
he
Peo
ple
Torc
hes
Ind
ief0
6452b
3e2
9336a3964e2
24b
f88a2f0
3
85
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
So
Col
dF
urt
her
See
ms
Fore
ver
Pen
ny
Bla
ckIn
die
33d
a9e7
63f7
7ed
ff5fd
f863099fa
2621
IG
otP
lenty
o’N
utt
inE
lla
Fit
zger
ald
an
d...
Ell
a&
Lou
isS
ing..
.Jazz
3c9
919393ee
96f4
965650ad
350b
1d
14d
Ingr
edia
nts
Con
stel
lati
onof
Cars
Con
stel
lati
on
of
...
Ind
ieb
9e5
9d
7e1
523fd
83ae4
6a34487fa
a3d
e
Let
’sC
all
Th
eW
hol
...
Ell
aF
itzg
erald
an
d...
Th
eB
est
Of
Ell
a..
.Jazz
f88299b
27722e6
c33961210d
736f6
5cf
Lim
ehou
seB
lues
Dja
ngo
Rei
nh
ard
tT
he
Bes
tof
Dja
ng..
.Jazz
9fe
ab
b3d
f172cc
f7fc
bca
1507ab
6fd
02
Moch
aS
pic
eE
mil
yR
emle
rR
etro
spec
tive
Vol.
..Jazz
c3b
1d
cd0798e7
d4b
bca
f14e2
4aa50d
24
My
Mel
anch
oly
Bab
yD
izzy
Gil
lesp
ieJazz
Coll
ecto
rE
d..
.Jazz
d6f4
f32fb
5fe
236e0
e527a8486d
9ea
58
Sev
enC
ome
Ele
ven
Ch
arli
eC
hri
stia
nT
he
Ori
gin
al
Gu
it..
.Jazz
14d
523e
f6c3
bf1
328543d
6525ed
2d
a54
So
Mu
chfo
rth
eA
ft..
.E
verc
lear
So
Mu
chfo
rth
eA
...
Alt
ern
ati
ve01b
0d
2564c5
5fe
9d
31e9
a1eb
551aff
b4
Son
gF
rom
anA
mer
ic..
.E
verc
lear
Son
gs
Fro
man
Am
e...
Alt
ern
ati
ve47e0
7a7b
faf8
478fb
02b
ac8
2c5
a0a3c2
Sp
arks
Are
Gon
na
Fly
Cat
her
ine
Wh
eel
Wis
hvil
leA
lter
nati
ve18d
5eb
152c6
2d
67c2
792c1
64386a0405
Su
per
vix
enG
arb
age
Garb
age
Alt
ern
ati
vefe
c9605433fa
2f5
15e0
3a92fc
db
76668
Th
eR
eap
ing
Coh
eed
and
Cam
bri
aG
ood
Ap
oll
oI’
mB
...
Pu
nk
7a3f9
ebf5
eb0ca
4e5
74a32d
6e0
792d
62
Th
eR
ing
inR
etu
rnC
ohee
dan
dC
am
bri
aIn
Kee
pin
gS
ecre
t...
Pu
nk
d1b
9b
40a8c2
c61e0
0b
9ff
d3f0
9d
1f0
4e
You
Mu
stB
eL
osin
g..
.F
ats
Wal
ler
Port
rait
Vol.
2-
...
Jazz
ae1
b4b
92e0
9d
3570043ee
70285c7
db
86
You
rG
eniu
sH
and
sE
verc
lear
Worl
dO
fN
ois
eA
lter
nati
ve5a8e8
9d
901ed
723e9
bafb
12d
961f0
87f
Squ
are
On
eC
old
pla
yX
&Y
Ind
ieb
ca69d
1e5
03a266b
e0d
ff0aad
0d
dd244
Not
My
Fu
ner
alC
hil
dre
nof
Bod
om
Rel
entl
ess,
Rec
kl.
..M
etal
ab
ba7b
0cc
2804f7
5550ea
343197fb
752
Bed
Bra
nd
New
Dais
yP
un
k29090ef
eac1
ab
3ab
4ab
39562fd
7b
836b
Bes
tof
Me
Big
wig
Un
-Mer
ryM
elod
ies
Pu
nk
11664ce
ded
60fb
bf2
57b
fa787155003d
86
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Dir
tR
oad
Blu
esB
obD
yla
nT
ime
Ou
tO
fM
ind
Folk
3fe
a99180d
86004ae1
b65264816f1
bd
6
Cap
rico
rnB
raid
Fra
nkie
Wel
fare
B..
.In
die
12e6
2d
edaa1277fd
6b
c2c7
678d
54a3a4
Jim
my
Go
Sw
imm
erB
raid
Mov
ieM
usi
cvo
l.T
wo
Ind
ie94e6
bf3
e0148fa
f11e4
17b
5d
72100e8
d
Cou
ntr
yH
ouse
Blu
rT
he
Gre
at
Esc
ap
eA
lter
nati
ve74316b
f48a634751c8
a48b
b4b
d3e7
cb7
Day
ligh
tB
ette
rT
han
Ezr
aB
efore
Th
eR
ob
ots
Alt
ern
ati
ve19d
38b
c9ce
464e4
94d
46f5
190001f8
67
Dec
kT
he
Hal
lsB
ing
Cro
sby
Fra
nk
...
Ch
rist
mas
Jazz
08f9
3cd
3d
bfa
7f4
bab
7023d
9cb
bf2
49d
Den
tB
igw
igS
tay
Asl
eep
Pu
nk
cfb
b431d
fed
942ea
47802d
24ae2
017d
a
Kra
cked
Din
osau
rJr.
You
’re
Liv
ing
All
...
Alt
ern
ati
vef4
b04a9
69ca
e0456ed
613e7
49eff
b1fb
Don
’tC
han
geY
our
P..
.B
enF
old
sF
ive
Th
eU
nau
thori
zed
...
Ind
ie0c2
77b
6498f9
129e8
492e6
8168e4
61ae
Fai
rB
enF
old
sF
ive
Wh
ate
ver
An
dE
ver.
..In
die
cbaf3
01821d
15c1
c68c8
480554141ec
8
Fel
icia
Blu
esT
ravel
erS
traig
ht
on
Til
l..
.A
lter
nati
ve6e8
a17042e4
8d
3b
ae6
98343ea
27137cc
Fu
ture
Boy
Cat
her
ine
Wh
eel
Ad
am
An
dE
ve-
Me.
..A
lter
nati
vece
377ca
5b
ef874d
e53f4
08632ed
5d
10d
Good
Bet
ter
Th
anE
zra
Del
uxe
Alt
ern
ati
ve7899ed
cf51b
5ce
7b
003b
e4d
13b
f10532
Hig
her
Pow
erB
osto
nG
reate
stH
its
Rock
4488cb
3b
572f9
4039a186d
45c2
83cd
ee
IC
onfe
ssC
ath
erin
eW
hee
lC
hro
me
Alt
ern
ati
vee1
77893b
b9a483f0
565b
a0852b
1f5
1fe
IW
ant
To
Tou
chY
ouC
ath
erin
eW
hee
lF
erm
ent
Alt
ern
ati
veb
c5b
7c0
cafa
596b
b5b
900ea
2593e9
dc2
Lon
ely
Man
Bil
lF
rise
llD
isfa
rmer
(Sta
nd
ard
)Jazz
18a880d
fc2984d
5d
8047ce
0d
3c7
78af2
Lov
e&
Gre
edB
lues
Tra
vel
erS
ave
His
Sou
lA
lter
nati
ve5c8
cbd
014c3
216f0
42e1
d5e9
47ee
5214
Mon
key
Wre
nch
Foo
Fig
hte
rsC
olo
ur
&T
he
Sh
ap
eA
lter
nati
ve9b
e9f0
17a186d
d574a3b
754ec
86f7
f67
Mot
ion
Lig
ht
Bra
idM
ovie
Mu
sic
Vol.
On
eIn
die
bf6
090b
6cf
69155d
609762d
bd
1539d
6a
87
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Old
Tim
esB
ill
Fri
sell
and
85..
.S
ign
Of
Lif
eJazz
f6369a74e7
3f7
9666d
fc7887f3
ab
9705
Qu
een
Of
Sor
row
Bla
ckL
abel
Soci
ety
Han
gov
erM
usi
cV
o..
.M
etal
9b
bf8
c5e2
dd
0b
cccf
4ee
461d
90e5
d84a
Rew
ind
Bet
ter
Th
anE
zra
Fri
ctio
n,
Baby
Alt
ern
ati
vea4e7
d24
51493fc
41262a01afb
28b
0e1
6
Son
g2
Blu
rB
lur
Alt
ern
ati
ve33ca
9cc
3b
897611d
5789e7
23b
0f5
9725
Sta
nd
Blu
esT
ravel
erF
ou
rA
lter
nati
ve9e1
9cf
2fd
3ad
3e8
67a126c8
837f3
c809
Sw
imB
ush
Six
teen
Sto
ne
Alt
ern
ati
vefa
f033c9
395a268348e1
f1b
1ad
3d
54b
3
take
ach
ance
onlo
veB
rian
Set
zer
13
Rock
abil
ly29ab
dc9
f2889c6
9fb
f672d
76cd
5cb
9c8
Th
eN
ewS
tyle
Bea
stie
Boy
sL
icen
sed
To
Ill
Hip
-Hop
ba6f7
75d
dcb
0d
a7e6
7b
e1a4f0
1b
6c0
8b
Th
eR
ose
Pet
alle
dG
...
Bla
ckL
abel
Soci
ety
Son
icB
rew
Met
al
1875ff
8c5
c03601d
7d
2049b
2a6a203f4
Tou
ghG
uy
Bea
stie
Boy
sIl
lC
om
mu
nic
ati
on
Hip
-Hop
a4d
c9f7
b5219aa5e5
74d
0e7
f5c6
b7ce
1
Way
dow
nC
ath
erin
eW
hee
lH
ap
py
Day
sA
lter
nati
ve13ae2
6883ab
12a4d
2f5
37e6
dd
3363190
Wh
at’s
InY
ouB
lack
Lab
elS
oci
ety
Mafi
aM
etal
b3e2
95d
2178b
695e8
9d
15ee
127fc
20eb
Wh
ere
My
For
tun
eL
ies
Cac
oph
ony
Sp
eed
Met
al
Sym
ph
ony
Met
al
774b
f6669892c8
70e7
ea0011c6
52b
dcd
Cra
nk
(Liv
e)C
ath
erin
eW
hee
lB
roke
nN
ose
[Dis
c1]
Alt
ern
ati
ve37b
da82a2b
808f6
90f8
052c8
8d
498c2
f
Del
icio
us
Cat
her
ine
Wh
eel
Ma
Soli
tud
a[D
isc
1]
Alt
ern
ati
vec3
4d
a38
4e6
9fb
0e7
4822231b
f8c7
07cc
Flo
wer
To
Hid
e(L
iv..
.C
ath
erin
eW
hee
lB
roke
nN
ose
[Dis
c2]
Alt
ern
ati
vee5
ae7
2d
82b
eb65504a9ed
6670d
0fb
cd8
Gir
lF
rom
Th
eN
orth
...
Bob
Dyla
nT
he
Fre
ewh
eeli
n’
...
Folk
d2c7
a66
a637809a192c9
fa4b
3046c3
88
God
Insi
de
My
Hea
dC
ath
erin
eW
hee
lJu
dy
Sta
rin
gA
tT
...
Alt
ern
ati
veb
844139
5999a057d
561b
f5332f4
54605
Kil
lin
gA
Cam
era
Bra
idF
ram
e&
Canva
sIn
die
7f7
301b
d8c5
cc0ad
720b
561ad
84c0
057
Log
anT
oG
over
nm
ent.
..B
ran
dN
ewT
he
Holi
day
EP
Pu
nk
5d
a68c8
04775aa41c5
4d
2e4
a607e5
4e7
88
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Nin
etee
n75
Bra
idT
he
Age
of
Oct
een
Ind
ie33252b
29264ad
52ed
752b
a888f9
1790b
Ple
asu
reC
ath
erin
eW
hee
lC
ran
kA
lter
nati
ve542a82676d
c2ea
0b
3e7
954cb
3e7
86f9
2
Th
eN
ewN
ath
anD
etr.
..B
raid
Lu
cky
To
Be
Ali
veIn
die
1f3
a9d
c3a53974176ca
167d
ba00d
b90a
Up
sid
eD
own
Cat
her
ine
Wh
eel
Sh
e’s
My
Fri
end
(Ep
)A
lter
nati
ve0c5
cd89
d1f0
1f4
050a6984f7
30c3
753f
Urs
aM
ajo
rS
pac
eSt.
..C
ath
erin
eW
hee
lI
Want
To
Tou
chY
ou
Alt
ern
ati
ve57c0
bf9
63563ad
dec
e43c3
fe0f8
634f2
Wis
hY
ouW
ere
Her
eC
ath
erin
eW
hee
lL
ike
Cats
An
dD
ogs
Alt
ern
ati
ve15e7
dc5
b93fa
ceef
a498fa
26ea
53f4
d3
AF
oggy
Day
Bil
lie
Hol
iday
Bil
lie’
sB
est
Jazz
3e2
ca0d
5a5e2
b276412b
0c3
840301318
Bal
lad
Of
Hol
lis
Bro
wn
Bob
Dyla
nT
he
Tim
esT
hey
Ar.
..F
olk
125d
437
978169d
38a91d
b3948266ed
76
Cal
low
ayB
oog
ieB
igB
adV
ood
oo
Dad
dy
How
Big
Can
You
Get
Sw
ing
b4a60e9
f2d
faf1
c9980f0
2c3
ac2
8ea
64
Bla
ckC
row
Blu
esB
obD
yla
nA
noth
erS
ide
of
B..
.F
olk
d6d
27d
3441111c1
7903aff
45b
46ee
fdd
Tra
cyJac
ks
Blu
rP
ark
life
Alt
ern
ati
ve02f0
bff
9e4
ea3f9
ff0e5
1b
ebf7
5c1
347
Do
Over
Bra
idC
lose
rT
oC
lose
dIn
die
dd
f40b
64d
1e1
b80b
33033a979f4
5ab
81
IW
ann
aB
eL
ike
You
Big
Bad
Vood
oo
Dad
dy
Th
isB
eau
tifu
lL
ife
Sw
ing
9704fc
e21d
c958b
d8ca
24e8
3a3c3
4f0
3
Isis
Bob
Dyla
nD
esir
eF
olk
6f5
2ad
d5316a5600b
166a7f5
0ab
0b
77f
Lou
isia
na
Bar
ney
Kes
sel
To
Sw
ing
or
Not
t...
Jazz
b0c8
5e2
3f7
313fb
8c7
52d
5d
3f2
863115
Mr.
Pin
stri
pe
Su
itB
igB
adV
ood
oo
Dad
dy
Big
Bad
Vood
oo
Dad
dy
Sw
ing
ba6b
b2d
d5805918836b
0e9
e27cc
df7
34
Nas
hvil
leS
kyli
ne
Rag
Bob
Dyla
nN
ash
vil
leS
kyli
ne
Folk
576b
d7d
0e1
ad
c88d
8613fc
4898279965
Ple
dgi
ng
My
Tim
eB
obD
yla
nB
lon
de
on
Blo
nd
eF
olk
1fa
d92f
9e3
d4760a4d
0ab
a138b
b47c2
1
Sat
inD
oll
Bar
ney
Kes
sel
Th
eP
oll
Win
ner
sJazz
b579ef
af8
6ff
89c9
f7624f8
bb
48d
a885
Sh
eb
elon
gsto
me
Bob
Dyla
nB
rin
gin
gIt
All
B..
.F
olk
22b
7242
7876fb
a473570e1
079e2
a5377
89
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Sim
ple
Tw
ist
Of
Fat
eB
obD
yla
nB
lood
On
Th
eT
rack
sF
olk
d3266ca
04492026531d
dcc
9fd
5a5e2
16
Th
ese
Fool
ish
Th
ings
Ben
ny
Good
man
Ben
ny
Good
man
Jazz
a2b
fd0898943e0
023d
32492ae6
be7
4eb
Tim
eA
fter
Tim
eB
rian
Nov
aS
had
owO
fY
ou
rS
mil
eJazz
5c6
72114ef
76e9
2aa990a7867fd
0a76e
Tom
bst
one
Blu
esB
obD
yla
nH
ighw
ay61
Rev
isit
edF
olk
9ef
29e8
d38d
4d
54ee
e3cf
a11ea
d12850
Tw
enty
Yea
rsB
ill
Fri
sell
wit
hD
...
Bil
lF
rise
llw
ith
...
Jazz
18cc
8b
ea220b
8260e6
f7b
d27460eb
537
Ver
ona
Bil
lF
rise
llG
on
e,Ju
stL
ike
a..
.Jazz
4b
c872c
50cd
e31d
d2f9
0b
25f4
a3e9
bde
Vol
are
Bar
ney
Kes
sel
Th
eP
oll
Win
ner
s..
.Jazz
91c4
524c3
5d
ca0893b
111fb
0ab
245764
Wh
atA
reY
ouD
oin
g..
.B
arn
eyK
esse
lS
olo
Jazz
c6b
3836
562cd
e9619cd
935f6
5e6
ab
794
You
Kn
owY
ouW
ron
gB
igB
adV
ood
oo
Dad
dy
Sav
eM
yS
ou
lS
win
gf6
ab
963
a2e2
ad
6c7
bb
c617b
f53d
b801c
Sic
Tra
nsi
tG
lori
a...
.B
ran
dN
ewD
eja
Ente
ndu
Pu
nk
c13b
0d
274d
cd77d
6eb
c0421076c3
4a35
(Inte
rmis
sion
)B
ane
ItA
llC
om
esD
own
...
Pu
nk
55ca
42b
4b
f4125707e0
307b
d6e1
ec3d
f
AB
reak
,A
Pau
seA
sT
all
As
Lio
ns
As
Tall
As
Lio
ns
Ind
ie4b
ae0
3972d
8173b
5b
a2625a5c1
865fc
2
Ain
tL
ove
Gra
nd
Atr
eyu
Su
icid
eN
ote
san
d..
.M
etal
340f1
9b
946306d
17b
075fb
64cc
fcb
746
An
dth
eW
ick
Bu
rnt
...
As
Tal
lA
sL
ion
sB
lood
An
dA
ph
ori
sms
Ind
ie8ad
8b
b13d
54d
39856e1
128839e1
b799b
Tea
rsF
orT
he
Sh
eep
Atm
osp
her
eL
ucy
Ford
,T
he
At.
..H
ip-H
op
7463405b
5842a4b
eefe
e0d
8d
fa92b
435
Th
eS
kin
ny
Atm
osp
her
eW
hen
Lif
eG
ives
Y..
.H
ip-H
op
990988ac6
f20306e9
9f5
389678237000
Au
gust
Ava
ilO
ver
Th
eJam
esP
un
kd
32c3
bcf
8a1276f2
9b
6f6
42fe
6e8
8640
Lar
edo
Ban
dO
fH
orse
sIn
fin
ite
Arm
sIn
die
99644d
04520b
20871d
370b
6039a5e6
36
Bew
ild
erB
adly
Dra
wn
Boy
Th
eH
ou
rO
fB
ewil
...
Ind
ie6732556976b
41cc
5b
bb
ef5b
251966408
Bir
dS
ings
Why
Th
e..
.A
tmos
ph
ere
Sev
en’s
Tra
vel
sH
ip-H
op
cad
ffc5
f20ac0
73187d
39c1
69ec
c5c0
d
90
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Bot
hG
un
sB
lazi
ng
Ban
eH
old
ing
Th
isM
om
ent
Pu
nk
10273ad
1f0
6602358d
7d
a06f3
8d
5e2
af
Can
’tS
top
ItB
adR
elig
ion
Th
eP
roce
ssO
fB
e...
Pu
nk
0491823d
d23c8
3042447f8
eb0c5
5d
258
Car
ryM
eH
ome
Atm
osp
her
eS
ad
Clo
wn
Bad
Sp
r...
Hip
-Hop
3b
6f5
6b929fb
e9b
e5ee
37deb
f34d
154a
Cri
cifi
edA
uto
mat
ic7
Au
tom
ati
c7
Pu
nk
51b
6ef
ecd
60f2
80887d
08a49c0
3c9
dc2
Cu
rren
tS
tatu
sA
tmos
ph
ere
Ove
rcast
!H
ip-H
op
d3aee
9aa01d
e99949c1
91cc
97cb
30281
Em
pir
eG
rad
eA
ud
iocr
ush
So
You
Call
Th
ese.
..In
die
7e9
f4a08a85b
678c4
1574280d
0b
68399
Giv
eM
eA
tmos
ph
ere
God
Lov
esU
gly
Hip
-Hop
e299516fb
2c5
3f1
9e3
def
88b
d848921e
How
Mu
chIs
En
ough
Bad
Rel
igio
nS
uff
erP
un
k3ad
ec87
758303140197ed
bb
801c8
cd87
N30
Ava
ilO
ne
Wre
nch
Pu
nk
eef9
5f8
cc3a72147b
49204c4
b0ce
54ad
No
Dir
ecti
onB
adR
elig
ion
Gen
erato
rP
un
k809704371972ee
f1c6
9e1
fbcf
d0ad
7b
c
No
On
e’s
Gon
na
Lov
e...
Ban
dof
Hor
ses
Cea
seto
Beg
inIn
die
69284508d
dd
1a1907ca
27ff
f1cf
b11b
a
On
eA
rmed
Sci
ssor
At
Th
eD
rive
-In
Rel
ati
on
ship
Of
C..
.P
un
kf3
216f4
bc2
0fc
23f4
c18715739696c1
7
Ou
rS
wor
ds
Ban
dO
fH
orse
sE
very
thin
gA
llT
h..
.In
die
9c0
9108527cc
ceec
090b
10ef
e85321b
1
Sin
gle
Bad
Ast
ron
au
tH
ou
ston
:W
eH
ave
...
Pu
nk
c568800d
5b
68f5
89c8
0874e7
96d
14c0
8
Sn
akes
Am
ong
Us
Ban
eG
ive
Blo
od
Pu
nk
fe3883d
f6ee
fc26a8a9683487c0
cac6
d
Sor
eL
oser
sB
igw
igA
nIn
vit
ati
on
To
...
Pu
nk
d0d
74ab
7101c0
ac1
c70145b
06e7
bcb
a5
Sou
nd
Of
AG
un
Au
dio
slav
eR
evel
ati
on
sA
lter
nati
vea0d
63d
b1321c5
30fc
620ea
3c2
023c5
d9
Str
ange
rT
han
Fic
tion
Bad
Rel
igio
nS
tran
ger
Th
an
Fic
...
Pu
nk
35b
6938
5d
b8216401a5a1b
0658ee
aa86
Th
eyA
llG
etM
adA
t...
Atm
osp
her
eS
ad
Clo
wn
Bad
Win
...
Hip
-Hop
e6d
608e
7d
90f4
2580d
edf7
69d
ae5
83f6
Tor
ture
sO
fT
he
Dam
ned
Bay
sid
eB
aysi
de
Pu
nk
e77b
b5c5
238d
9487ff
8f4
535329a8831
91
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Tu
esd
ayA
vail
4A
MF
rid
ayP
un
k9cc
afc
04fd
05923d
ff55f6
a9200cc
df1
Tu
nin
gA
vail
Dix
ieP
un
k40a8d
ed9d
f15e3
38ca
cd70ef
57e9
87d
7
Urs
aM
inor
At
Th
eD
rive
-In
VA
YA
Pu
nk
c868fa
a105e3
ea9ee
68f3
f9d
592e8
c0e
RF
TC
Atm
osp
her
eS
ad
Clo
wn
Bad
Su
m..
.H
ip-H
op
ef85a6b
c3ce
70aad
c15ad
197b
2c8
32aa
Wat
chO
ut
Atm
osp
her
eY
ou
Can
’tIm
agin
e...
Hip
-Hop
40383eb
3d
691c4
55d
385b
322a8220775
Bec
ame
Atm
osp
her
eT
he
Fam
ily
Sig
nH
ip-H
op
092d
afa
fee7
fb65843c2
1b
a2b
503e1
4b
Ju
stIn
Tim
eB
arn
eyK
esse
lL
et’s
Cook!
Jazz
0983f0
f4f8
033ca
8b
4d
27391b
d2955a1
My
Rev
erie
Bar
ney
Kes
sel
Mu
sic
To
Lis
ten
To
Jazz
a070a0f7
661ea
0d
0702a37746f3
a4289
On
aS
low
Boa
tto
C..
.B
arn
eyK
esse
lK
esse
lP
lays
Sta
n..
.Jazz
17b
1a07
6f1
ce9a42123fd
265f3
b39cc
1
Poor
Bu
tter
fly
Bar
ney
Kes
sel
Poor
Bu
tter
fly
Jazz
bc7
6e2
768b
a5787614aaf5
2c8
f9c3
083
Th
eL
ittl
eR
hu
mb
aB
arn
eyK
esse
lP
oll
Win
ner
sT
hre
eJazz
1a332e9
4c7
6633496a902e0
28203e8
52
You
Can
’tT
ake
ItW
...
As
Tal
lA
sL
ion
sY
ou
Can
’tT
ake
It..
.In
die
381c8
d428e6
03451499b
e4b
75368fc
a1
Sca
lpA
tmos
ph
ere
To
All
My
Fri
end
s...
Hip
-Hop
d26b
8d
ad
a3489fe
6611536cb
42934444
Anti
soci
alA
nth
rax
Sta
teO
fE
up
hori
aM
etal
446634261b
64a37f1
1d
b266637e0
d798
Ble
edT
he
Fre
akA
lice
InC
hai
ns
Face
lift
Alt
ern
ati
ve39185a4d
0ca
09b
460ce
db
b4a5930d
73f
Con
tin
enta
lA
lkal
ine
Tri
oG
ood
Mou
rnin
gP
un
k8c9
19d
22f6
5886e9
86ed
d1b
cb08c4
43a
Dea
dO
nT
he
Flo
orA
lkal
ine
Tri
oT
his
Ad
dic
tion
Pu
nk
0b
bd
692514093e4
271a2e3
57b
37210e0
Dow
nIn
AH
ole
Ali
ceIn
Ch
ain
sD
irt
Alt
ern
ati
ve634a0e5
eb7094f8
52ed
0b
c46916ca
b0a
Efi
lnik
ufe
sin
(N.F
.L.)
Anth
rax
Am
on
gT
he
Liv
ing
Met
al
ad
d816c0
c4aa06075d
69aab
efce
ec6c2
F**
*Y
ouA
uro
raA
lkal
ine
Tri
oM
ayb
eI’
llC
atc
h..
.P
un
k975a6b
89f5
62d
7ff
0e0
bae7
8c8
8ce
b98
92
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Fu
nn
els
All
anH
old
swort
hA
llN
ight
Wro
ng
Jazz
c5fd
f8b
afa
c24522d
6992ab
13796acb
6
Hea
ven
Bes
ide
You
Ali
ceIn
Ch
ain
sA
lice
InC
hain
sA
lter
nati
ve3092cd
fb02e0
eafc
081afe
538fe
bcc
dc
IL
ied
My
Fac
eO
ffA
lkal
ine
Tri
oA
lkali
ne
Tri
oP
un
ke7
8883f7
cb4ee
257419e8
6eb
cafe
ba6d
Mer
cyM
eA
lkal
ine
Tri
oC
rim
son
Pu
nk
bd
06f6
ba5fb
ac6
e8fa
66c8
6f8
53c7
483
No
Excu
ses
Ali
ceIn
Ch
ain
sJar
Of
Fli
esA
lter
nati
ve596ec
c48411f8
20a65b
9d
bb
3ed
7ce
a1b
Nos
eO
ver
Tai
lA
lkal
ine
Tri
oG
od
dam
nit
Pu
nk
223f3
5478d
c7674d
c8b
401af5
e847c8
4
Slu
dge
Fac
tory
Ali
ceIn
Ch
ain
sM
TV
Un
plu
gged
Alt
ern
ati
ve8ec
0a73f3
122c4
a52ea
f82cc
e7a0256b
Sta
min
aA
llE
lse
Fai
lsT
he
Ch
ase
Of
Gain
Pu
nk
01cb
3a2
1ac4
a211b
df5
839d
2e4
80a225
Stu
pid
Kid
Alk
alin
eT
rio
Fro
mH
ere
toIn
fi..
.P
un
k20d
356a
4a20f9
d7713acc
8d
c61fd
8536
Th
eC
arou
sel
As
Tal
lA
sL
ion
sL
afc
ad
ioIn
die
2e8
a24929fc
d2647c0
4d
976e1
7e7
c15d
Tok
yoD
ream
All
anH
old
swort
hR
oad
Gam
esJazz
f78b
a49
afb
41d
3ef
5d
6039e1
5f6
a3fe
1
Tu
nin
gA
vail
Liv
eat
the
Bott
o..
.P
un
k48f0
29b
7ef
de3
4fa
94fa
25b
4943a798a
Cap
acit
yT
oC
han
geC
ath
erin
eW
hee
lJu
dy
Sta
rin
gA
tt.
..A
lter
nati
ve667d
70b
9413b
eef7
87667898d
c87ec
0a
Eve
ryT
hu
gN
eed
sA
...
Alk
alin
eT
rio
Dam
nes
iaP
un
k680ed
e279047d
4a5e4
07f9
f3f0
f89cb
2
Ove
rA
nd
Ou
tA
lkal
ine
Tri
oA
gony
An
dIr
ony
Pu
nk
31af0
53af9
09a5ed
b5d
2f8
f526a25b
7e
Sto
p!
Aga
inst
Me!
New
Wav
eP
un
k17c3
ebc0
e889d
205ca
22aa94d
35e5
e03
Bot
hF
eet
InA
lex
Skol
nic
kT
rio
Tra
nsf
orm
ati
on
Jazz
5c9
ad
b4d
a2487d
302fb
a11b
3d
a4e5
531
Ch
emic
alP
arty
Gav
inD
eGra
wC
hari
ot
Pop
Rock
577d
a2e
958ac0
f2f8
4990a0b
18b
c2a3a
Des
troy
Wh
atD
estr
o...
Aga
inst
All
Au
thori
tyD
estr
oyW
hat
Des
t...
Pu
nk
7d
9f9
549c6
e6775643fd
329987420e3
5
Let
ter
from
Ind
iaA
lD
iM
eola
,P
aco
D..
.T
he
Gu
itar
Tri
oJazz
12b
3ab
ece0
f2c6
f7e8
a90ea
349140596
93
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Mov
ing
Mol
ecule
sA
gain
stT
omorr
ow’s
Sky
Th
eL
ost
Tap
esIn
die
14337afe
fd7d
fbab
996f6
d4b
ab
3608d
3
Per
fect
ion
All
Mass
Ner
der
Pu
nk
5cc
39c4
f4afc
e0fd
ed852457a2094213
Slu
rrin
gth
eR
hyth
ms
Aga
inst
Me!
As
Th
eE
tern
al
Co..
.P
un
kd
769a05
bc7
a39725429ce
164940f6
1ad
Tak
ing
Not
esA
gain
stT
omorr
ow’s
Sky
Ju
mp
Th
eH
edges
F..
.In
die
07610b
2e5
46c0
fc622af4
333b
772104c
Fro
mH
erL
ips
To
Go.
..A
gain
stM
e!S
earc
hin
gfo
ra
F..
.P
un
kb
933759
5041c0
84c2
feb
a7a8b
7d
04964
Scr
eam
ItU
nti
lY
ou..
.A
gain
stM
e!R
einve
nti
ng
Axl
Rose
Pu
nk
67c4
658ea
ec4fd
721a5569c8
f8fd
7a92
Gir
l’s
Not
Gre
yA
FI
Sin
gT
he
Sorr
owP
un
k7d
bc6
f9f5
1f4
b660795d
c06b
07fd
71f7
Sta
nd
InL
ine
Aga
inst
All
Au
thori
tyA
llF
all
Dow
nP
un
k1d
ef858
db
1301b
15e9
377a8a5d
8e9
a2b
Th
eC
hec
kere
dD
emon
AF
IA
nsw
erT
hat
An
dS
...
Pu
nk
82eb
7ae
9f1
7b
cbf1
4f3
c2366a32e0
4c2
Th
eR
ipA
Wil
hel
mS
crea
mM
ute
Pri
nt
Pu
nk
9864e4
1a64ee
a592a025a12cc
5d
b9f2
1
Me
Vs.
Mor
riss
eyIn
...
AW
ilh
elm
Scr
eam
Ru
iner
Pu
nk
b3c0
59a
dc3
fc5163a0e8
21c3
bafd
bc2
a
My
Lit
tle
Wor
ld88
Fin
gers
Lou
ieB
ehin
dB
ars
Pu
nk
0ec
2c6
fbd
488058125a064564927f2
80
Th
eP
raye
rP
osit
ion
AF
IB
lack
Sail
sIn
Th
...
Pu
nk
a08464d
82d
1c0
4b
b86a430668b
59e1
21
Th
eory
Of
Rev
olu
tion
AF
IV
ery
Pro
ud
Of
Ya
Pu
nk
3aa7d
d13069847c3
b312fa
ea3606b
c84
Sle
epin
gB
eau
tyA
Per
fect
Cir
cle
Mer
de
Nom
sM
etal
a1ac9
410542acf
e936cc
e30a8797c7
0e
Get
Mad
,Y
ouS
onof
...
AW
ilh
elm
Scr
eam
Care
erS
uic
ide
Pu
nk
9d
39915
37c1
3d
7456c0
88727d
ad
9b
e88
Th
eC
rew
7S
econ
ds
Th
eC
rew
Pu
nk
379e2
be9
c252608a31b
083004b
98600a
Rel
ativ
eW
ays
...A
nd
You
Wil
lK
no..
.S
ou
rce
Tags
&C
od
esIn
die
be0
3b
3d
8f3
60fa
3fa
e53a1137394658e
You
rsp
irit
’sal
ive
Dro
pkic
kM
urp
hys
Th
eW
arr
ior’
sC
od
eP
un
kb
dc7
f9fe
909c7
419454e2
f1cf
6e9
d195
Ice
Mac
hin
eD
epec
he
Mod
eS
pea
k&
Sp
ell
New
Wav
e91b
b30ca
028456ca
fa4e5
6aeb
c9494c4
94
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
IfY
ouW
ant
(Liv
ei.
..D
epec
he
Mod
eS
om
eG
reat
Rew
ard
New
Wav
e3a3625f6
14956af7
04f4
a4c1
f1a8d
7a5
My
Sec
ret
Gar
den
(L..
.D
epec
he
Mod
eA
Bro
ken
Fra
me
New
Wav
e0d
27382
498312ae8
5aad
b5778407e8
8b
Som
eth
ing
toD
oD
epec
he
Mod
eS
om
eG
reat
Rew
ard
New
Wav
ef8
4f7
cb12e0
768ee
a752b
2f1
250544c5
Th
eG
reat
Ou
tdoor
s!D
epec
he
Mod
eC
on
stru
ctio
nT
ime.
..N
ewW
ave
1595849c9
364ef
b974765af4
ed25d
311
Wal
kA
way
Dro
pkic
kM
urp
hys
Bla
ckout
Pu
nk
a5ac7
bfd
d4cc
c7a0b
8f5
60a7e1
5aa963
Sco
ttis
hW
ind
sF
righ
ten
edR
ab
bit
AF
righte
ned
Rab
b..
.In
die
b98c3
6369867ab
e3a05627ef
55224f9
1
We’
reB
reak
ing
Up
Aga
inst
Me!
Wh
ite
Cro
sses
Pu
nk
3934d
ef3e5
fb05d
35e4
ae5
81358d
cc74
Lif
elin
e(L
ive)
Cat
her
ine
Wh
eel
Wis
hvil
le(B
onu
s..
.A
lter
nati
vec2
8214a2cb
48a33b
af8
ca1c1
419816f6
Del
icio
us
(Liv
e)C
ath
erin
eW
hee
lM
aS
oli
tud
a[D
isc
2]
Alt
ern
ati
vead
65cc
dd
3aff
c7a5b
8a6d
27a96e5
633e
20th
Cen
tury
Tow
ers
Dea
thC
abfo
rC
uti
eT
he
Sta
bil
ity
EP
Ind
ie886f0
d39e7
5b
3334200c2
e3ed
867ff
3f
Mak
esth
eS
un
Com
eO
ut
Atm
osp
her
eS
ad
Clo
wn
Bad
Fal.
..H
ip-H
op
d972cd
8fc
95a917d
13ab
0f8
5086c1
e47
Fu
llM
oon
Atm
osp
her
eS
tric
tly
Lea
kage
Hip
-Hop
96ca
9225f6
99c2
0b
fbce
53cb
a562fc
ce
Th
eR
opes
Atm
osp
her
eL
eak
At
Wil
lH
ip-H
op
b1b
9b
f2c3
b10d
228d
4ee
9b
7d
536496c3
Ju
de
Law
An
dA
Sem
e...
Bra
nd
New
You
rF
avori
teW
ea..
.P
un
k85b
b505acf
410c4
7994d
43c4
33b
f0fa
b
By
My
Sid
eC
opel
and
Eat,
Sle
ep,
Rep
eat
Ind
ieff
b0157a
b1af8
4b
d09d
346f1
422ef
3d
a
Mar
chin
gB
and
sof
M..
.D
eath
Cab
for
Cu
tie
Pla
ns
Ind
ie9d
05518
bd
5418826eb
3e2
7d
e3f5
b1b
62
Wor
ldIn
My
Eyes
Dep
ech
eM
od
eV
iola
tor
New
Wav
e2cc
56cf
ceb
0d
a75477ee
e1eb
f7639d
9a
ID
on’t
Th
ink
Din
osau
rJr
Han
dIt
Ove
rA
lter
nati
ve800e5
3e1
02c5
dd
1c5
c5421b
55047ea
f9
Eve
nY
ouD
inos
aur
Jr
Wit
hou
ta
Soun
dA
lter
nati
ve9f2
d333
976d
ec167d
6fd
d7ff
bae4
3c6
8
Ou
tT
her
eD
inos
aur
Jr.
Wh
ere
You
Bee
nA
lter
nati
vecd
64872
3346a4e0
8330b
26b
f1744ea
81
95
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Day
One
Exp
losi
ons
InT
he
Sky
Tra
vels
InC
on
sta..
.In
die
db
e72d
22ac5
3d
00d
587e7
6d
d4b
0487c6
Hid
den
Tra
ckB
ran
dN
ewT
he
Dev
ilA
nd
God
...
Pu
nk
eb24454
91602c0
45a5a9688f9
45a5929
Bee
tleb
um
Blu
rB
lur
Alt
ern
ati
vefa
c5ca
2618f4
c77609c9
27c0
7e6
f434f
Lov
eS
ick
Bob
Dyla
nT
ime
Ou
tO
fM
ind
Folk
55e0
8b
ad
e8946059d
8c2
b9ef
0a494060
Bor
edT
oT
ears
Bla
ckL
abel
Soci
ety
Son
icB
rew
Met
al
9c9
a648843b
f6b
0ea
ab
4313548b
a72c2
An
gel
Fal
lsB
raid
Fra
nkie
Wel
fare
B..
.In
die
5eb
d9e3
36b
72d
b92f9
2813160a91d
89e
Ele
ph
ant
Bra
idM
ovie
Mu
sic
vol.
Tw
oIn
die
9f2
36988233f0
44f6
01277b
f342d
4b
45
Bu
rned
Bet
ter
Th
anE
zra
Bef
ore
Th
eR
ob
ots
Alt
ern
ati
ve1c2
c018a01b
878736400b
889d
337b
a9a
Car
olin
aB
lues
Blu
esT
ravel
erS
traig
ht
on
Til
l..
.A
lter
nati
vea42ff
aea
b9a4e8
a1c7
ff12a70ca
ac5
a1
Cra
zyO
rH
igh
Bla
ckL
abel
Soci
ety
Han
gov
erM
usi
cV
o..
.M
etal
904b
00f
be6
f2a1eb
c9c8
793b
105eb
f78
Dis
farm
erT
hem
eB
ill
Fri
sell
Dis
farm
er(S
tan
dard
)Jazz
d0851e1
98cf
58a6e4
2f2
85fe
b55eb
cf4
Dol
lF
oo
Fig
hte
rsC
olo
ur
&T
he
Sh
ap
eA
lter
nati
veb
eafa
a7cc
e81925c0
4583a09c2
a0e0
57
dru
gsan
dal
coh
ol(.
..B
rian
Set
zer
13
Rock
abil
ly254ee
d1fa
2e0
7e4
2d
c840d
4ad
5d
0aed
b
Eve
ryth
ing
Zen
Bu
shS
ixte
enS
ton
eA
lter
nati
ve1d
b566692ab
782878a6f8
b436b
a930fa
Fir
eIt
Up
Bla
ckL
abel
Soci
ety
Mafi
aM
etal
3032f9
5b
41b
f245a3b
62b
fc66f8
28161
God
Insi
de
My
Hea
dC
ath
erin
eW
hee
lH
ap
py
Day
sA
lter
nati
vec7
8a496a20276e6
cd01862fa
3b
4ff
bd
5
InT
he
Blo
od
Bet
ter
Th
anE
zra
Del
uxe
Alt
ern
ati
ve2e4
9c9
81220459333b
6fa
ac4
b4ac3
2e7
Intr
oC
ath
erin
eW
hee
lA
dam
An
dE
ve-
Me.
..A
lter
nati
ve90105b
95c4
1e5
67d
3ce
f29732b
13f4
80
It’s
AL
ong
Sto
ryB
ill
Fri
sell
and
85..
.S
ign
Of
Lif
eJazz
6b
497b
483d
b4e8
cbaf2
3fd
52e3
db
96cf
Jin
gle
Bel
lsB
ing
Cro
sby
Fra
nk
...
Ch
rist
mas
Jazz
84d
1122
713ac2
48b
dd
440aef
a8b
027ac
96
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Kil
lR
hyth
mC
ath
erin
eW
hee
lC
hro
me
Alt
ern
ati
ve01eb
135
a36b
5b
da25ea
6443d
d8c4
847a
Kin
gO
fN
ewO
rlea
ns
Bet
ter
Th
anE
zra
Fri
ctio
n,
Baby
Alt
ern
ati
ve2d
85132
134f1
0d
c4a2ef
d7d
bd
88ea
223
Nar
cole
psy
Ben
Fol
ds
Fiv
eT
he
Un
au
thori
zed
...
Ind
iefa
b8789
e67f0
23e8
fd71a696c2
049d
56
Old
Lad
yB
igw
igU
n-M
erry
Mel
od
ies
Pu
nk
c6cf
7c7
0d
c137c7
094f6
e73c9
2101fd
d
On
eA
ngr
yD
war
fA
nd
...
Ben
Fol
ds
Fiv
eW
hate
ver
An
dE
ver.
..In
die
2b
1d
a5185c7
c7fe
70d
4c4
be1
6e6
2e0
f6
Rhym
in&
Ste
alin
Bea
stie
Boy
sL
icen
sed
To
Ill
Hip
-Hop
021216c9
6669c0
cb818865f9
a6108962
Ru
n-A
rou
nd
Blu
esT
ravel
erF
ou
rA
lter
nati
ved
267fd
3e6
a876a756b
ae3
522ef
2954f4
Sav
age
Cac
oph
ony
Sp
eed
Met
al
Sym
ph
ony
Met
al
91ca
a9608f2
1444378cd
13322ab
d8692
Sou
nd
sL
ike
Vio
len
ceB
raid
Mov
ieM
usi
cV
ol.
On
eIn
die
22a9095ed
b9986657f0
029d
4ad
c9b
c9b
Ste
reot
yp
esB
lur
Th
eG
reat
Esc
ap
eA
lter
nati
ve3f0
e1029f6
93d
9601754afe
19d
978685
Sti
llB
igw
igS
tay
Asl
eep
Pu
nk
888484c9
cb973e5
87eb
e69206f6
0a1d
8
Su
reS
hot
Bea
stie
Boy
sIl
lC
om
mu
nic
ati
on
Hip
-Hop
bc2
760f
6aa1a6a5e1
88453a011758524
Tel
lM
eB
osto
nG
reate
stH
its
Rock
26e1
6008421974c2
9d
01ca
263b
86f1
de
Tex
ture
Cat
her
ine
Wh
eel
Fer
men
tA
lter
nati
ve6e0
a8e5
8514c8
9ed
ae9
df4
15e7
c2c7
9e
Tri
na
Mag
na
Blu
esT
ravel
erS
ave
His
Sou
lA
lter
nati
ve55b
d0f0
40b
5ae4
f079e6
8e8
096e5
6462
Vic
esB
ran
dN
ewD
ais
yP
un
k2b
db
917e8
9f4
bc1
5d
4b
37e6
d7b
3c8
72a
Au
tob
iogr
aphy
Bra
idL
uck
yT
oB
eA
live
Ind
ie0a87b
c5863a3964c5
6855c2
17cd
674d
e
Blo
win
’In
Th
eW
ind
Bob
Dyla
nT
he
Fre
ewh
eeli
n’
...
Folk
8b
fa9fe
28150e8
1883b
1a7675d
5a60f1
Bro
ken
Nos
e(L
ive
Xfm
)C
ath
erin
eW
hee
lB
roke
nN
ose
[Dis
c2]
Alt
ern
ati
ve9c9
d534
7a003e4
3e7
9c9
2b
e48171195a
Bro
ken
Nos
e(S
ingl
e...
Cat
her
ine
Wh
eel
Bro
ken
Nose
[Dis
c1]
Alt
ern
ati
vee8
f6160b
f992cc
3ea
2456a2c7
5d
cfd
69
97
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Cra
nk
Cat
her
ine
Wh
eel
Cra
nk
Alt
ern
ati
ve6d
97a63
e6a61cc
9b
8ee
0b
cc0e9
871b
0b
Hea
l2
Cat
her
ine
Wh
eel
Lik
eC
ats
An
dD
ogs
Alt
ern
ati
vea5101389f4
2d
0b
3e5
48d
b53d
6e5
bd
840
IW
ant
To
Tou
chY
ouC
ath
erin
eW
hee
lI
Want
To
Tou
chY
ou
Alt
ern
ati
vec9
45ec
def
a8d
d72248c9
0792754d
cab
1
Ju
dy
Sta
rin
gA
tT
he.
..C
ath
erin
eW
hee
lJu
dy
Sta
rin
gA
tT
...
Alt
ern
ati
ve51e0
e2af2
ae3
99d
043d
2f5
bc5
67ab
157
Ma
Sol
itu
da
Cat
her
ine
Wh
eel
Ma
Soli
tud
a[D
isc
1]
Alt
ern
ati
vee8
69f9
b2ed
164758777ea
a823c3
826c1
My
Bab
yS
mok
esB
raid
Th
eA
ge
of
Oct
een
Ind
ie981113f6
bce
1a0ee
5fe
706569f9
0e5
2e
Th
eB
oyW
ho
Blo
cked
...
Bra
nd
New
Th
eH
oli
day
EP
Pu
nk
003f7
b27ab
972166f3
164747d
9fa
a9d
b
Th
eN
ewN
ath
anD
etr.
..B
raid
Fra
me
&C
anva
sIn
die
560a7d
fac4
4589265261982d
4d
3e4
ea7
All
IR
eall
yW
ant
t...
Bob
Dyla
nA
noth
erS
ide
of
B..
.F
olk
a5d
c53a
7d
f87ed
122b
1a49e8
2b
37ce
e4
Be
Dee
dle
De
Do
Bar
ney
Kes
sel
Th
eP
oll
Win
ner
s..
.Jazz
b601c4
f90d
94262b
978b
87ae3
c4f0
c58
Beg
inth
eB
lues
Bar
ney
Kes
sel
To
Sw
ing
or
Not
t...
Jazz
f098ea
f10ca
0d
6142d
1cc
9053cb
29d
0f
Big
An
dB
adB
igB
adV
ood
oo
Dad
dy
Th
isB
eau
tifu
lL
ife
Sw
ing
2d
1ce
aa809d
3969c4
938550342e5
8e6
8
Com
eO
nW
ith
Th
e”c
...
Big
Bad
Vood
oo
Dad
dy
How
Big
Can
You
Get
Sw
ing
4194ef
67d
ab
865407f6
ad
92cf
c8a6784
Blu
esfo
rL
osA
nge
les
Bil
lF
rise
llG
on
e,Ju
stL
ike
a..
.Jazz
e7d
8065
f07b
c6a4ef
53e2
c7465ee
293d
Gir
ls&
Boy
sB
lur
Park
life
Alt
ern
ati
ve3e4
b2cc
ead
1cf
6ce
2c1
6afe
aa78c8
3cd
Th
eR
ight
Tim
eB
raid
Clo
ser
To
Clo
sed
Ind
ie8e2
8ca
263790b
965428cd
59362b
f00eb
Bra
zil
Bar
ney
Kes
sel
Solo
Jazz
d36d
e4eb
6e9
0b
7c8
245f8
5fc
812e0
238
Gir
lF
rom
Th
eN
orth
...
Bob
Dyla
nN
ash
vil
leS
kyli
ne
Folk
88a53710e1
e8d
b6e9
929817fb
1c5
e0b
7
Hu
rric
ane
Bob
Dyla
nD
esir
eF
olk
bb
ba56
226b
d4c7
083126f5
be0
640aaff
Jor
du
Bar
ney
Kes
sel
Th
eP
oll
Win
ner
sJazz
550e8
ea3c3
ce4fa
23e2
930b
0c5
ecb
4cb
98
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Let
’sD
ance
Ben
ny
Good
man
Ben
ny
Good
man
Jazz
cb48b
927b
58661b
28192d
2f3
fbf0
0f0
c
Lik
eA
Rol
lin
gS
ton
eB
obD
yla
nH
ighw
ay61
Rev
isit
edF
olk
9b
38e1
1212b
5d
f8095691b
22c5
2b
c9a4
Ou
tlaw
sB
ill
Fri
sell
wit
hD
...
Bil
lF
rise
llw
ith
...
Jazz
c07ef
027168b
d5cf
c42275d
e3c8
83851
Rai
ny
Day
Wom
en12
...
Bob
Dyla
nB
lon
de
on
Blo
nd
eF
olk
bea
2ad
16e2
35904ab
d835afa
e6b
bcd
f1
Su
bte
rran
ean
hom
esi.
..B
obD
yla
nB
rin
gin
gIt
All
B..
.F
olk
d9d
9f7
1c9
3b
23fb
242b
169279ce
5b
557
Tan
gled
Up
InB
lue
Bob
Dyla
nB
lood
On
Th
eT
rack
sF
olk
1e9
f981e0
33ca
604705863e4
eafc
41f9
Th
eB
oog
ieB
um
per
Big
Bad
Vood
oo
Dad
dy
Big
Bad
Vood
oo
Dad
dy
Sw
ing
f381b
78177016f7
b01b
49393422b
bf4
5
Th
eS
had
owO
fY
our
...
Bri
anN
ova
Sh
ad
owO
fY
ou
rS
mil
eJazz
9e0
c2f9
6b
d5d
69f4
fa7a581ac4
2f8
1b
9
Th
eT
imes
Th
eyA
re...
Bob
Dyla
nT
he
Tim
esT
hey
Ar.
..F
olk
19b
c1a4
aeb
2d
572b
5924cd
0ef
b8a6d
f3
Wh
atA
Lit
tle
Moon
l...
Bil
lie
Hol
iday
Bil
lie’
sB
est
Jazz
553a5e6
e98fb
86e8
49b
7ce
7c3
d61eb
cc
Zig
Zag
gity
Woop
Wo.
..B
igB
adV
ood
oo
Dad
dy
Sav
eM
yS
ou
lS
win
g52c3
ac3
98ad
ef617139b
99a338a73eb
9
Tau
tou
Bra
nd
New
Dej
aE
nte
ndu
Pu
nk
441ec
43b
52888c5
26c2
b037a5ce
9d
b88
1000
Mor
eF
ool
sB
adR
elig
ion
Su
ffer
Pu
nk
0b
c21a4
ee8d
83af1
32b
223b
e094d
2170
66th
Str
eet
Atm
osp
her
eS
ad
Clo
wn
Bad
Win
...
Hip
-Hop
84e3
05c0
72b
049690cd
548048f8
ca361
Lik
eT
od
ayA
tmos
ph
ere
Lu
cyF
ord
,T
he
At.
..H
ip-H
op
d436b
127d
433d
9d
37c6
a3124a03f9
345
Pu
pp
ets
Atm
osp
her
eW
hen
Lif
eG
ives
Y..
.H
ip-H
op
1373cd
e6b
aae9
c371f8
90c0
2a0551d
f0
Com
pli
men
tsB
and
Of
Hor
ses
Infi
nit
eA
rms
Ind
ie216d
d2eb
1636ff
3924554e7
73b
2ec
9f4
Bri
efD
escr
ipti
onA
tmos
ph
ere
Ove
rcast
!H
ip-H
op
c3d
9542
b89a73f6
45204d
d5061d
6d
3a0
Cle
arC
utt
ing
Bad
Ast
ron
au
tH
ou
ston
:W
eH
ave
...
Pu
nk
22b
b4ed
61d
e116d
31c4
f6544ed
965728
Clo
ne
Ava
ilD
ixie
Pu
nk
b9f2
8b
357b
5fd
cf1816655ff
2b
6c8
9ad
99
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Cou
nt
Me
Ou
tB
ane
Hold
ing
Th
isM
om
ent
Pu
nk
b3f8
9d
e5482e0
520a8d
f561f2
9182951
Dan
cin
gin
the
Rea
r...
As
Tal
lA
sL
ion
sB
lood
An
dA
ph
ori
sms
Ind
ieb
c294c9
e12743fc
03aa7aeb
81c3
5855e
Dev
otio
n&
Des
ire
Bay
sid
eB
aysi
de
Pu
nk
f7e1
3b
88acc
62b
c3c1
8227e2
bc1
e8501
Dil
ated
Atr
eyu
Su
icid
eN
ote
san
d..
.M
etal
840b
589
669d
ec3869eb
d79c7
b8d
65fd
b
Eve
ryb
od
y’s
Sta
lkin
gB
adly
Dra
wn
Boy
Th
eH
ou
rO
fB
ewil
...
Ind
ie78b
090e
3c9
5ec
a34022b
ed39b
8544459
Good
Dad
dy
Atm
osp
her
eS
ad
Clo
wn
Bad
Sp
r...
Hip
-Hop
432e3
df7
2d
a43853875a96cf
35e8
cef1
InH
ere
Au
dio
cru
shS
oY
ou
Call
Th
ese.
..In
die
b6fb
017d
9500c9
175b
a9ad
d98c8
3f4
76
Lea
veM
ine
To
Me
Bad
Rel
igio
nS
tran
ger
Th
an
Fic
...
Pu
nk
446ca
c33288717ab
43d
4e7
8a99925c5
1
Lu
cky
One
Au
tom
atic
7A
uto
mati
c7
Pu
nk
04f2
b2a
28d
5f9
466b
911c8
11402a4b
a5
New
2A
vail
Ove
rT
he
Jam
esP
un
ke9
4b
691
8cb
41d
5301d
ea151e7
4b
8155c
Od
eto
LR
CB
and
ofH
orse
sC
ease
toB
egin
Ind
ie819aa4e0
b6e1
7a2a89ed
e2f3
11fa
cb37
On
eA
nd
Th
eS
ame
Au
dio
slav
eR
evel
ati
on
sA
lter
nati
veec
e47c1
6b
22b
11d
2108728237a5a7cf
7
Ord
erA
vail
4A
MF
rid
ayP
un
kc3
31947c0
a20c5
360b
bd
2d
52e7
f68162
Pat
tern
Aga
inst
US
erA
tT
he
Dri
ve-I
nR
elati
on
ship
Of
C..
.P
un
kc1
fdaa9
8d
f4f2
e937b
3b
4a797781d
958
Pro
veIt
Bad
Rel
igio
nT
he
Pro
cess
Of
Be.
..P
un
k1f1
db
fdb
85ec
b964c3
d3175521b
bef
40
Pro
xim
aC
enta
uri
At
Th
eD
rive
-In
VA
YA
Pu
nk
b67fc
f3d
0e8
5a8d
f4a0b
7506b
c18b
d79
Sin
kO
rSw
imB
igw
igA
nIn
vit
ati
on
To
...
Pu
nk
507314ec
50b
048d
b7ac8
396b
dff
ba684
Som
eC
ame
Ru
nn
ing
Ban
eG
ive
Blo
od
Pu
nk
9d
52079
fed
ca997663b
2d
191d
4b
2ca
4e
Son
gfo
rL
un
aA
sT
all
As
Lio
ns
As
Tall
As
Lio
ns
Ind
ie2f9
6fe
e705d
1e1
5fd
c3247f0
7f5
4b
9f4
Str
ide
Ava
ilL
ive
at
the
Bott
o..
.P
un
kcd
4368e
b7afb
54d
54ad
f6222c0
782002
100
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Str
uck
Dow
nB
yM
eB
ane
ItA
llC
om
esD
own
...
Pu
nk
fb10289
7b
71404b
8ca
8d
f29b
6b
7a7643
Tak
enA
vail
On
eW
ren
chP
un
k33c4
4924f4
a199757d
fc8f6
795194a6d
Th
eB
ass
An
dT
he
Mo.
..A
tmos
ph
ere
God
Lov
esU
gly
Hip
-Hop
4ec
82a8b
ed38c6
2ca
59313a1d
53fd
005
Too
Mu
chT
oA
skB
adR
elig
ion
Gen
erato
rP
un
k9b
e103c
5ac6
41557a5b
8e1
85ec
ca788b
Try
ing
To
Fin
dA
Ba.
..A
tmos
ph
ere
Sev
en’s
Tra
vel
sH
ip-H
op
380cd
a82743629d
fdb
72905665ef
50b
5
Wic
ked
Gil
Ban
dO
fH
orse
sE
very
thin
gA
llT
h..
.In
die
020b
278
09b
4ab
87ea
22ee
50759568a9f
Dea
dG
irlf
rien
dC
ath
erin
eW
hee
lI
Want
toT
ou
chY
...
Alt
ern
ati
vee4
fa54b
58c7
a3238563b
5b
b220e5
f50b
Pan
icA
ttac
kA
tmos
ph
ere
You
Can
’tIm
agin
e...
Hip
-Hop
0f5
b379
ba96993ef
69ad
bfe
592f1
4f2
9
Th
eN
um
ber
On
eA
tmos
ph
ere
Sad
Clo
wn
Bad
Su
m..
.H
ip-H
op
5f9
854426ec
0c2
399340a9ab
74783f3
3
Th
eL
ast
To
Say
Atm
osp
her
eT
he
Fam
ily
Sig
nH
ip-H
op
7ce
feca
0e2
f73d
8e5
a17d
f1d
12fc
3152
Cri
sis
Bar
ney
Kes
sel
Poll
Win
ner
sT
hre
eJazz
865e8
0f2
a4239fb
b43f1
8cf
392fb
5ca
4
Lov
eis
Her
eto
Sta
yB
arn
eyK
esse
lK
esse
lP
lays
Sta
n..
.Jazz
8b
901ed
a9ad
a1d
c6b
7c7
c9a8c3
6a4c1
8
Mak
in’
Wh
oop
eeB
arn
eyK
esse
lM
usi
cT
oL
iste
nT
oJazz
40e9
54e0
bff
dce
9b
51120a8e3
02d
af0
0
Mon
sieu
rA
rman
dB
arn
eyK
esse
lP
oor
Bu
tter
fly
Jazz
ec117afa
8aff
2963ee
744484b
039ad
f8
Six
es&
Sev
ens
As
Tal
lA
sL
ion
sY
ou
Can
’tT
ake
It..
.In
die
a363e4
31590fd
873b
859876a56fd
8387
Tim
eR
emem
ber
edB
arn
eyK
esse
lL
et’s
Cook!
Jazz
12b
e2ea
42d
93ac9
4f0
565c4
caa02f9
b8
Th
eM
ajo
rL
eagu
esA
tmos
ph
ere
To
All
My
Fri
end
s...
Hip
-Hop
2ff
279a832b
e106af8
ced
5b
712668276
Alp
hra
zall
anA
llan
Hol
dsw
ort
hA
llN
ight
Wro
ng
Jazz
722eb
5ab
b14d
02c1
842010d
8219b
36d
a
Ble
eder
Alk
alin
eT
rio
Alk
ali
ne
Tri
oP
un
k1d
29b
345b
7ec
14a82b
e0eb
ad
db
1d
f58b
Bu
rnA
lkal
ine
Tri
oC
rim
son
Pu
nk
cfcf
c4c9
6782c7
4cc
ba01ab
a7ad
19b
c9
101
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
IA
mT
he
Law
Anth
rax
Am
on
gT
he
Liv
ing
Met
al
fce2
f471b
b3123ff
c214c0
8f5
d20f1
7d
IS
tay
Aw
ayA
lice
InC
hai
ns
Jar
Of
Fli
esA
lter
nati
veab
266fe
1c2
78e8
50707c7
55ec
f2b
3d
ff
Lea
dP
oiso
nin
gA
lkal
ine
Tri
oT
his
Ad
dic
tion
Pu
nk
719ac2
3ff
b162b
6fc
111501cb
85b
caaa
Mak
eM
eL
augh
Anth
rax
Sta
teO
fE
up
hori
aM
etal
0afa
c457b
2797c1
609e6
c50ee
0d
54510
No
Excu
ses
Ali
ceIn
Ch
ain
sM
TV
Un
plu
gged
Alt
ern
ati
ve20781943b
6289b
0f5
98ce
2613d
49c4
26
On
eH
un
dre
dSto
ries
Alk
alin
eT
rio
Good
Mou
rnin
gP
un
k5b
a8d
67666f6
572613577d
5a5582a997
Rai
nW
hen
ID
ieA
lice
InC
hai
ns
Dir
tA
lter
nati
ve94918f3
b6434b
2c9
f17b
9239c3
1b
5f6
3
San
Fra
nci
sco
Alk
alin
eT
rio
God
dam
nit
Pu
nk
baaf7
31a8d
4314eb
28fb
9c1
5e3
ec44b
9
Sea
Of
Sor
row
Ali
ceIn
Ch
ain
sF
ace
lift
Alt
ern
ati
vec3
b64cc
4e0
92e2
d7292af9
03c3
b67001
Sil
hou
ette
sS
ilh
oue.
..A
sT
all
As
Lio
ns
Lafc
ad
ioIn
die
305f9
6412d
2fe
d1c8
f4408ca
c145a9d
6
Slu
dge
Fac
tory
Ali
ceIn
Ch
ain
sA
lice
InC
hain
sA
lter
nati
vefe
1603c1
535ed
f112ec
d470c4
dc9
df6
f
Tak
eL
ots
Wit
hA
lcoh
olA
lkal
ine
Tri
oF
rom
Her
eto
Infi
...
Pu
nk
827d
6a2
4b
7066ae8
706f3
fd032a9cff
0
Wat
eron
the
Bra
in,.
..A
llan
Hol
dsw
ort
hR
oad
Gam
esJazz
dd
9c8
fe67b
65034b
6b
f2c8
82597c1
536
Win
dfa
llA
llE
lse
Fai
lsT
he
Ch
ase
Of
Gain
Pu
nk
dcf
c270
1b
a5399f3
93d
795f1
47e0
51a2
You
’ve
Got
So
Far
T..
.A
lkal
ine
Tri
oM
ayb
eI’
llC
atc
h..
.P
un
k30a58a0ab
3d
9050504fa
6c0
20352d
2b
b
Gli
tter
Cat
her
ine
Wh
eel
Ju
dy
Sta
rin
gA
tT
...
Alt
ern
ati
ve20812f9
5c5
841d
3b
9d
b232b
bb
8ed
3d
62
Roof
top
sA
lkal
ine
Tri
oA
lkali
ne
Tri
o,
Ho..
.P
un
k79a0b
3c1f1
fba642c1
ccd
fb813a682d
6
Th
isC
ould
Be
Lov
eA
lkal
ine
Tri
oD
am
nes
iaP
un
kc8
a624559cc
c4447cc
fd06310d
6b
6506
InV
ein
Alk
alin
eT
rio
Agony
An
dIr
ony
Pu
nk
6968b
5067f4
bc6
113125fc
6892d
16a9f
Wh
ite
Peo
ple
For
Pea
ceA
gain
stM
e!N
ewW
ave
Pu
nk
b23d
e3d
cde5
572f0
561cc
ffa4814a734
102
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Can
dy
InM
yC
avit
yA
gain
stT
omorr
ow’s
Sky
Th
eL
ost
Tap
esIn
die
fae6
e8d
de7
caa135865ea
acc
d429e6
03
Fai
rwea
ther
Fri
end
All
Mass
Ner
der
Pu
nk
523ff
ad
9536ef
a23c9
eca173aa245427
Fre
edom
Aga
inst
All
Au
thori
tyD
estr
oyW
hat
Des
t...
Pu
nk
377d
571
3c0
607128379479d
e31b
d6b
1c
Man
ha
De
Car
nav
alA
lD
iM
eola
,P
aco
D..
.T
he
Gu
itar
Tri
oJazz
2fd
b6d
4a3b
2f7
a025fc
48f0
1278a7632
Mon
eyA
lex
Skol
nic
kT
rio
Tra
nsf
orm
ati
on
Jazz
b4892fb
835b
c2c6
483832b
3d
9eb
78d
4b
Pen
Aga
inst
Tom
orr
ow’s
Sky
Ju
mp
Th
eH
edges
F..
.In
die
d5b
3ef
9065a0eb
20363ce
600d
bc5
6073
Sin
kF
lori
da
Sin
kA
gain
stM
e!A
sT
he
Ete
rnal
Co..
.P
un
k5f9
9b
415641b
76b
563165fa
4b
515d
b4d
IS
till
Lov
eY
ouJu
lie
Aga
inst
Me!
Rei
nve
nti
ng
Axl
Rose
Pu
nk
1f9
ed4c
dd
9a3255049ee
cb93b
71e0
ca0
Un
pro
tect
edS
exW
it..
.A
gain
stM
e!S
earc
hin
gfo
ra
F..
.P
un
k100d
3d
56ad
3017d
041d
b85336d
ffb
b3c
At
Ou
tE
xp
ense
Aga
inst
All
Au
thori
tyA
llF
all
Dow
nP
un
k677b
428
172cc
1d
257992b
7d
402f4
31a0
Bra
nd
New
Me,
Sam
e..
.A
Wil
hel
mS
crea
mM
ute
Pri
nt
Pu
nk
12095afb
bd
e72d
74271d
be2
2a71c2
153
Bro
wn
ieB
otto
mS
un
dae
AF
IA
nsw
erT
hat
An
dS
...
Pu
nk
224243fb
93cb
e83328619a244a8306a3
Dan
cin
gT
hro
ugh
Su
nd
ayA
FI
Sin
gT
he
Sorr
owP
un
k91a3ff
935795894fc
415139645914a83
Ad
van
ces
InM
od
ern
...
AF
IV
ery
Pro
ud
Of
Ya
Pu
nk
5d
98360
3ab
038b
067d
beb
4135e1
87c7
9
Clo
veS
mok
eC
ath
arsi
sA
FI
Bla
ckS
ail
sIn
Th
...
Pu
nk
5fe
d7b
77d
4ad
7cf
49cb
dd
1d
de7
cbb
d55
Hol
din
gB
ack
88F
inge
rsL
ou
ieB
ehin
dB
ars
Pu
nk
cda74f4
b621c4
93d
b305636ab
c9b
e541
InV
ino
Ver
itas
IIA
Wil
hel
mS
crea
mR
uin
erP
un
k3d
1d
1969ff
084f7
929a0f0
06b
98d
556b
3L
ibra
sA
Per
fect
Cir
cle
Mer
de
Nom
sM
etal
00f4
b93
acb
387c1
93f7
412d
e88fd
ead
a
You
rL
ove
AW
ilh
elm
Scr
eam
AW
ilh
elm
Scr
eam
Pu
nk
5a8f1
ec97d
36011897a664b
79b
f927b
e
Th
ese
Dea
dS
tree
tsA
Wil
hel
mS
crea
mC
are
erS
uic
ide
Pu
nk
451b
630
010a23d
50ae6
934fe
341d
8e8
e
103
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Wh
atIf
Th
ere’
sa
W..
.7
Sec
ond
sT
he
Cre
wP
un
kb
60151e
fecf
14d
68208747c6
4a994e9
3
Day
sO
fB
ein
gW
ild
...A
nd
You
Wil
lK
no..
.S
ou
rce
Tags
&C
od
esIn
die
2798d
33a7a45d
1aff
423ed
a55d
5ff
e27
Bla
ckM
etal
lic
(Pee
...
Cat
her
ine
Wh
eel
Cra
nk
(12”)
Alt
ern
ati
ve655c1
6fe
433ff
c9e3
4c0
de4
29fc
8949d
Com
eB
ack
Aga
inC
ath
erin
eW
hee
lC
ran
k(C
ass
ette
)A
lter
nati
ve1cf
37c8
774a3256b
eef6
bc7
a18ef
98c2
Su
ffoca
tion
Aga
inst
Me!
Wh
ite
Cro
sses
Pu
nk
dcb
490cc
29c3
cf0ae1
d89fa
8a787c9
c8
Wh
atis
Th
isT
hin
g..
.C
har
lie
Par
ker
Yard
bir
dS
uit
eD
i...
Jazz
aa30fc
140f0
7d
a45997ad
b0d
9c6
fd0a4
Par
tyO
ver
Her
eA
tmos
ph
ere
Sad
Clo
wn
Bad
Fal.
..H
ip-H
op
75b
2886
fe2690433a9089ec
651838399
Ma
Sol
itu
da
(Tim
Fr.
..C
ath
erin
eW
hee
lM
aS
oli
tud
a[D
isc
2]
Alt
ern
ati
ve9001e0
f8a0ed
40fe
95fb
e478b
768144d
Lit
tle
Mat
hY
ouA
tmos
ph
ere
Str
ictl
yL
eaka
ge
Hip
-Hop
29d
51b
8e1
04e8
dd
ec58e6
1b
31493f5
b5
Th
eyA
lway
sK
now
Atm
osp
her
eL
eak
At
Wil
lH
ip-H
op
e08cc
d9389a9b
b63651e3
182184a4585
Th
eS
how
erS
cen
eB
ran
dN
ewY
ou
rF
avori
teW
ea..
.P
un
k014e3
f25799fd
3c4
62ec
fa5f3
10d
672b
Don
’tW
ant
toK
now
...
Cat
her
ine
Wh
eel
30th
Cen
tury
Man
...
Alt
ern
ati
ve532ce
1aa28b
5b
81b
b5405cc
15c0
b270e
1597
Atm
osp
her
eO
verc
ast
!H
ip-H
op
db
d078
9afa
80b
09e3
ca95b
9f9
25aa9d
9
AS
ong
For
Th
eO
pti
...
Atr
eyu
Su
icid
eN
ote
san
d..
.M
etal
b61d
01a0fe
83393b
ee1b
82e8
51afd
baa
Arc
arse
nal
At
Th
eD
rive
-In
Rel
ati
on
ship
Of
C..
.P
un
k160392d
90d
6944e6
d71ed
362e3
a733cd
Bet
wee
nT
he
Lin
esA
tmos
ph
ere
Lu
cyF
ord
,T
he
At.
..H
ip-H
op
60f0
4cd
9171d
8083b
1988841d
d02d
46d
Lik
eT
he
Res
tO
fU
sA
tmos
ph
ere
Wh
enL
ife
Giv
esY
...
Hip
-Hop
15f9
03b
3b
29b
d729ad
ef85ff
14d
e70b
9
Au
tum
nL
eave
sA
ud
iocr
ush
So
You
Call
Th
ese.
..In
die
981197fc
f544b
0943f8
3e4
10c4
207e3
4
Fac
tory
Ban
dO
fH
orse
sIn
fin
ite
Arm
sIn
die
47c1
aa01b
b6a5fd
51ef
4d
7f3
d2d
1d
56c
Bre
akB
loss
omA
sT
all
As
Lio
ns
Blo
od
An
dA
ph
ori
sms
Ind
ie46e3
11e9
657898aa897fe
82b
ad
5474ad
104
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Bu
ryM
eA
uto
mat
ic7
Au
tom
ati
c7
Pu
nk
9e8
d2cc
1071ee
5939cb
4b
837eb
f3603b
Dee
pw
ood
Ava
ilO
ver
Th
eJam
esP
un
k17411b
20cd
5b
7cc
52ee
7b
a071ff
cb91b
Don
’tS
top
Atm
osp
her
eS
ad
Clo
wn
Bad
Win
...
Hip
-Hop
cfb
2165
523e4
5e8
140e4
676e0
a3346ea
”Fas
tO
ne”
Ava
ilO
ne
Wre
nch
Pu
nk
4c4
a4a913773d
2919ff
f25ec
05ab
7546
F**
*W
hat
You
Hea
rdB
ane
ItA
llC
om
esD
own
...
Pu
nk
a04a9494b
ad
f8d
813c5
8b
e5d
0f9
1f9
1b
Gen
erat
orB
adR
elig
ion
Gen
erato
rP
un
kd
7ce
2f6
dd
3eb
ef38a40a32699fe
f0629
Hel
loS
***t
yB
aysi
de
Bay
sid
eP
un
k44b
8e6
a032277e2
0f4
03767b
78e6
3d
1f
His
tory
Atm
osp
her
eS
even
’sT
ravel
sH
ip-H
op
0678958fc
ad
aac3
5d
e7e3
7ad
83b
f2c9
6
InP
iece
sB
ane
Hold
ing
Th
isM
om
ent
Pu
nk
a7d
9cb
c8b
6f0
53fa
a812c4
50465058d
6
Inco
mp
lete
Bad
Rel
igio
nS
tran
ger
Th
an
Fic
...
Pu
nk
fa4793b
69d
a4c8
18d
81647b
62837e0
3f
IsT
her
ea
Gh
ost
Ban
dof
Hor
ses
Cea
seto
Beg
inIn
die
10cc
1a79c6
6783d
ca59e7
edf8
a1c6
b53
Les
sO
ne
Atm
osp
her
eS
ad
Clo
wn
Bad
Sp
r...
Hip
-Hop
62e3
bfc
ca01075b
684222c6
9184cd
e6c
On
Th
eN
od
Ava
ilD
ixie
Pu
nk
cda63cc
bcf
499288810a53c8
741f3
b16
On
emos
ph
ere
Atm
osp
her
eG
od
Lov
esU
gly
Hip
-Hop
f1b
201e
6d
4d
d27943b
42e3
b9462b
7d
07
Ras
cuac
he
At
Th
eD
rive
-In
VA
YA
Pu
nk
a66d
7d
10b
54fb
e6fc
995716b
04ac9
60c
Rev
elat
ion
sA
ud
iosl
ave
Rev
elati
on
sA
lter
nati
ved
12ee
a2d
15fc
666033099d
874b
086326
Sim
ple
Son
gA
vail
4A
MF
rid
ayP
un
k3ca
9e3
4877c2
e16759e5
2a5aa1a5cb
17
Sou
thB
oun
d95
Ava
ilL
ive
at
the
Bott
o..
.P
un
kca
f18aeb
33a2b
0b
4e7
6ace
f97d
f9ad
d4
Sp
eech
less
Ban
eG
ive
Blo
od
Pu
nk
373d
250
9025d
b1ee
3fa
fc007b
1e2
671b
Sta
bC
ity
As
Tal
lA
sL
ion
sA
sT
all
As
Lio
ns
Ind
ied
f17d
a444b
6aab
6cc
7173704b
af4
3872
105
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Su
per
son
icB
adR
elig
ion
Th
eP
roce
ssO
fB
e...
Pu
nk
ef66827396d
a576f1
9f2
d823e2
57f5
9f
Th
eF
irst
Son
gB
and
Of
Hor
ses
Eve
ryth
ing
All
Th
...
Ind
ieb
88f0
2f59ef
df3
12f0
e7b
e6856b
be6
58
Th
eS
hin
ing
Bad
lyD
raw
nB
oyT
he
Hou
rO
fB
ewil
...
Ind
ieed
17ed
7e9
388680c8
3064126ec
b47a43
Th
ese
Day
sB
adA
stro
nau
tH
ou
ston
:W
eH
ave
...
Pu
nk
dd
bb
3b
853b
8145699571e4
56d
0b
51d
7d
You
Are
(Th
eG
over
n..
.B
adR
elig
ion
Su
ffer
Pu
nk
8d
46566
665324a1d
7916044376fb
baa5
Su
nsh
ine
Atm
osp
her
eS
ad
Clo
wn
Bad
Su
m..
.H
ip-H
op
99e0
8b
d68fe
8fc
6571b
a35f3
08fc
c056
Th
eA
rriv
alA
tmos
ph
ere
You
Can
’tIm
agin
e...
Hip
-Hop
dee
e69e
23211e9
1e2
d3a8a5234aa5a1e
My
Key
Atm
osp
her
eT
he
Fam
ily
Sig
nH
ip-H
op
4d
d3c8
8f0
87b
47338cc
f750c3
27568e8
Ch
eerf
ul
Lit
tle
Ear
ful
Bar
ney
Kes
sel
Mu
sic
To
Lis
ten
To
Jazz
076f5
d0ed
0473834ac9
3b
b912a2e5
bc4
Cir
cles
As
Tal
lA
sL
ion
sY
ou
Can
’tT
ake
It..
.In
die
a42b
d09b
8a59074ff
ccd
264fb
334f7
6f
Dea
rly
Bel
oved
Bar
ney
Kes
sel
Poor
Bu
tter
fly
Jazz
eded
0e6
f16f5
7019b
79c2
a90cb
18c8
53
Let
’sC
ook
!B
arn
eyK
esse
lL
et’s
Cook!
Jazz
69b
18e7
102787ee
f6d
8f5
6338008b
b79
Sof
tW
ind
sB
arn
eyK
esse
lP
oll
Win
ner
sT
hre
eJazz
cfd
6873
b55fd
409a80f7
caa7850f6
c57
Sp
eak
Low
Bar
ney
Kes
sel
Kes
sel
Pla
ys
Sta
n..
.Jazz
21d
98b
f643cf
5ca
91e2
7c8
08c9
39fe
f2
Unti
lT
he
Nip
ple
sG
one
Atm
osp
her
eT
oA
llM
yF
rien
ds.
..H
ip-H
op
0d
25724
f531b
bb
a15b
4b
c2c7
587b
4f5
e
96H
eart
bea
tsA
sT
all
As
Lio
ns
Lafc
ad
ioIn
die
2fc
77b
7ae2
a81cf
04c7
3140d
8c2
fb8af
Anth
emA
llE
lse
Fai
lsT
he
Ch
ase
Of
Gain
Pu
nk
dea
6064
e0c2
25e7
59a5f3
1cf
db
78a2b
7
Bro
ther
Ali
ceIn
Ch
ain
sM
TV
Un
plu
gged
Alt
ern
ati
vea4c5
c2081fc
0fb
48e6
e669b
16b
289acf
Bru
shA
way
Ali
ceIn
Ch
ain
sA
lice
InC
hain
sA
lter
nati
ve4636a3b
d01b
7fe
8d
fa0a9e4
21a981d
6b
Cau
ght
InA
Mos
hA
nth
rax
Am
on
gT
he
Liv
ing
Met
al
52ae3
b75b
ea43d
c92eb
b4b
61e3
a2ca
0d
106
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Cop
Alk
alin
eT
rio
God
dam
nit
Pu
nk
297c7
a70182f7
6d
f112fc
f54a3a46f2
8
Dam
Th
atR
iver
Ali
ceIn
Ch
ain
sD
irt
Alt
ern
ati
ved
07b
b2104b
ae9
a0842793fe
376475623
Din
e,D
ine
My
Dar
lin
gA
lkal
ine
Tri
oT
his
Ad
dic
tion
Pu
nk
9e6
25766ce
14918a97b
c968f0
ea97c6
4
I’ll
Get
Th
ere
All
Mass
Ner
der
Pu
nk
58267e2
f9b
f2cb
f5b
395869ee
aff
4e6
3
Mad
amM
eA
lkal
ine
Tri
oM
ayb
eI’
llC
atc
h..
.P
un
kd
272554
ef8aad
db
7f8
9d
5d
3065b
6ca
4c
Man
InT
he
Box
Ali
ceIn
Ch
ain
sF
ace
lift
Alt
ern
ati
ve1b
e84cd
21857fb
eeec
cfb
31d
dae7
62b
6
Mr.
Ch
ain
saw
Alk
alin
eT
rio
Fro
mH
ere
toIn
fi..
.P
un
kfb
d1a7892a2624a56ed
260ec
6662c8
21
Nu
tsh
ell
Ali
ceIn
Ch
ain
sJar
Of
Fli
esA
lter
nati
ve5156fc
167c5
0301b
2a72f4
ff96014c5
3
Ou
tO
fS
ight,
Ou
tO
...
Anth
rax
Sta
teO
fE
up
hori
aM
etal
520a24f8
29b
d3cc
d27a1afe
144b
ce833
Roa
dG
ames
All
anH
old
swort
hR
oad
Gam
esJazz
6724cb
8fb
0a33b
8e6
8ac1
de0
f9418ce
2
Th
eP
oiso
nA
lkal
ine
Tri
oC
rim
son
Pu
nk
3b
988c3
54ea
505c8
63d
6f1
b3b
d45b
2d
f
Th
eT
hin
gsY
ouse
eA
llan
Hol
dsw
ort
hA
llN
ight
Wro
ng
Jazz
2ac2
594a3429a065fd
a2b
642acc
5f3
80
Th
isIs
Get
tin
gO
ve..
.A
lkal
ine
Tri
oA
lkali
ne
Tri
oP
un
k2700b
a179b
309fe
1e2
ba94cf
be4
4b
45d
We’
veH
adE
nou
ghA
lkal
ine
Tri
oG
ood
Mou
rnin
gP
un
k9d
bf6
3c2
d4253b
7b
3f2
fa86757c9
b1b
b
Nos
eO
ver
Tai
lA
lkal
ine
Tri
oD
am
nes
iaP
un
kea
fb4e7
d24f4
ff1b
dcb
652fb
0d
859d
44
Wh
ile
You
’re
Wai
tin
gA
lkal
ine
Tri
oA
lkali
ne
Tri
o,
Ho..
.P
un
kcc
52e8
2a4631b
58154e8
858d
9125e3
29
Hel
pM
eA
lkal
ine
Tri
oA
gony
An
dIr
ony
Pu
nk
7ed
ae0
48961160628051411ff
905b
8f3
Th
rash
Un
real
Aga
inst
Me!
New
Wav
eP
un
k6a7a523876a194fb
27e4
2d
9b
7e6
f6c9
0
Bli
nd
Fol
ds
and
Pad
...
Aga
inst
Tom
orr
ow’s
Sky
Th
eL
ost
Tap
esIn
die
8aec
ab
b9e0
faaee
1e6
0817557401d
0d
0
Con
dit
ion
ing
Aga
inst
All
Au
thori
tyD
estr
oyW
hat
Des
t...
Pu
nk
29b
c07a
e80e3
c1d
6e2
c6840ef
c879774
107
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Fea
rO
fF
lyin
gA
lex
Skol
nic
kT
rio
Tra
nsf
orm
ati
on
Jazz
5e1
1e9
d4c1
acd
6e7
b8aec
596c7
cd13d
6
Mid
sum
mer
Nig
ht
Al
Di
Meo
la,
Paco
D..
.T
he
Gu
itar
Tri
oJazz
2f5
ff4a0f4
f060ae9
79b
50f8
044ef
8e3
Mu
tiny
onth
eE
lecr
...
Aga
inst
Me!
As
Th
eE
tern
al
Co..
.P
un
kce
bb
9d
7165b
70fa
9d
2b
4c6
a3b
d21ed
82
Poi
son
Tes
ter
Aga
inst
Tom
orr
ow’s
Sky
Ju
mp
Th
eH
edges
F..
.In
die
3529f2
29c4
a62f8
ee11e6
6d
c9c1
16301
Ju
stin
Aga
inst
Me!
Sea
rch
ing
for
aF
...
Pu
nk
363ac4
23ef
e72af8
c2737233ea
f36883
We
Lau
ghat
Dan
ger
...
Aga
inst
Me!
Rei
nve
nti
ng
Axl
Rose
Pu
nk
56ef
99a3b
e326b
78e0
ab
82a96d
49c1
33
IW
ann
aG
etA
Moh
aw..
.A
FI
An
swer
Th
at
An
dS
...
Pu
nk
cc19d
2979f6
ecb
be0
2b
0d
923a38a6e7
b
Kee
pT
ryin
gA
gain
stA
llA
uth
ori
tyA
llF
all
Dow
nP
un
kc8
c706401e6
847fd
41c5
1a5c0
8f6
b5cf
Mal
leu
sM
alefi
caru
mA
FI
Bla
ckS
ail
sIn
Th
...
Pu
nk
8a24a1314ec
8a4d
e11c2
268a602ac3
4e
Sil
ver
An
dC
old
AF
IS
ing
Th
eS
orr
owP
un
k11ef
0a529533f5
29b
c32f5
381e9
88b
b7
Tot
alim
mor
tal
AF
IA
llH
all
ow’s
E.P
.P
un
k06579371f3
6fa
504ea
85b
8992a7c7
442
Wil
liam
sB
lake
Ove
r...
AW
ilh
elm
Scr
eam
Mu
teP
rint
Pu
nk
5169725b
bc9
a9b
b506fc
d5c5
99a6f5
d4
God
Lov
esa
Lia
rA
Wil
hel
mS
crea
mR
uin
erP
un
k6f2
a4f0
e65b
a70169588971e1
ae3
9fd
f
Jaw
s3,
Peo
ple
0A
Wil
hel
mS
crea
mC
are
erS
uic
ide
Pu
nk
0939170876214ab
753e1
a7c3
108a71a9
Per
fect
Fit
AF
IV
ery
Pro
ud
Of
Ya
Pu
nk
76d
de3
e86e0
811231ce
bb
e49492242e8
Sm
art
En
ough
To
Run
88F
inge
rsL
ou
ieB
ehin
dB
ars
Pu
nk
1e4
82cd
4504e4
526c9
b34893d
c771446
Ore
stes
AP
erfe
ctC
ircl
eM
erd
eN
om
sM
etal
5b
bd
795f4
1d
a562d
69625a19ec
94f3
3b
Skid
Rock
AW
ilh
elm
Scr
eam
AW
ilh
elm
Scr
eam
Pu
nk
c5a604536c5
6141a51ea
3d
65922d
46c6
Gir
lsB
east
ieB
oys
Lic
ense
dT
oIl
lH
ip-H
op
9d
01d
2a75ae6
ac9
ad
5c2
2fe
4f6
739d
4c
You
Los
e7
Sec
ond
sT
he
Cre
wP
un
k189728103b
a8837fc
950ad
668f2
8c3
33
108
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Mon
soon
...A
nd
You
Wil
lK
no..
.S
ou
rce
Tags
&C
od
esIn
die
54567a6cf
5829154674f9
1a6e9
6714b
b
Pey
ote
Atm
osp
her
eS
ad
Clo
wn
Bad
Fal.
..H
ip-H
op
956af7
1450300d
af0
43c7
6587b
079850
Bec
ause
Of
Th
eS
ham
eA
gain
stM
e!W
hit
eC
ross
esP
un
k3a73b
23f3
f6d
073724cd
54eb
13b
4285c
C’m
onA
tmos
ph
ere
Lea
kA
tW
ill
Hip
-Hop
e4783543f0
63640691d
728d
b47108571
Ygm
Atm
osp
her
eS
tric
tly
Lea
kage
Hip
-Hop
687f5
b59b
0fc
0ed
0d
879b
e292f4
31ac4
Am
ong
Th
eL
ivin
gA
nth
rax
Am
on
gT
he
Liv
ing
Met
al
ce7b
f99b
1d
0a96951b
91b
8b
f7135d
45b
Be
All
,E
nd
All
Anth
rax
Sta
teO
fE
up
hori
aM
etal
e63e5
2584945a19984f2
1828e6
e1d
fd0
Cri
nge
Alk
alin
eT
rio
God
dam
nit
Pu
nk
9fa
f97d
49f5
f144ee
df2
4b
42eb
c16fd
c
Div
isio
nA
llE
lse
Fai
lsT
he
Ch
ase
Of
Gain
Pu
nk
edd
e840b
88e1
9ea
cab
eda9c7
6e7
7fe
66
Good
bye
For
ever
Alk
alin
eT
rio
Alk
ali
ne
Tri
oP
un
k1974d
5407d
ebb
0a1b
d78f7
b0589ec
28d
Gri
nd
Ali
ceIn
Ch
ain
sA
lice
InC
hain
sA
lter
nati
ve761fd
5388cd
bf4
b221a4c6
0ed
40d
b0c9
Kee
p’E
mC
omin
gA
lkal
ine
Tri
oM
ayb
eI’
llC
atc
h..
.P
un
ka0eb
d3d
f9e0
83d
33acf
d95e2
65b
089d
b
Lan
yard
Loop
All
anH
old
swort
hA
llN
ight
Wro
ng
Jazz
0ac0
0cb
a14e3
cba0aad
3d
f1d
b47188ca
Nu
tsh
ell
Ali
ceIn
Ch
ain
sM
TV
Un
plu
gged
Alt
ern
ati
ve444b
538
4a3e5
25a51cc
a0b
791ec
04756
Pri
vate
Eye
Alk
alin
eT
rio
Fro
mH
ere
toIn
fi..
.P
un
k8b
cb0c2
8f0
a0f3
33b
846b
b44028643f1
Rot
ten
Ap
ple
Ali
ceIn
Ch
ain
sJar
Of
Fli
esA
lter
nati
ve1e5
65fb
418975ea
391b
a022374a6e5
9b
Th
emB
ones
Ali
ceIn
Ch
ain
sD
irt
Alt
ern
ati
ve13e0
1175265149b
58317ec
b4c1
53f5
ec
Th
isA
dd
icti
onA
lkal
ine
Tri
oT
his
Ad
dic
tion
Pu
nk
18857886c1
8d
f863b
2b
5d
f049b
54cf
3e
Th
isC
ould
Be
Lov
eA
lkal
ine
Tri
oG
ood
Mou
rnin
gP
un
kd
92f2
c30ef
6d
c2e4
82f0
02273aa3fe
44
Th
ree
Sh
eets
toth
e...
All
anH
old
swort
hR
oad
Gam
esJazz
92904a3d
ab
52ae4
af0
541546ef
bee
d95
109
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Tim
eT
oW
aste
Alk
alin
eT
rio
Cri
mso
nP
un
k549ce
3c8
365a908443478f0
e8b
ba1293
Was
teB
igw
igA
nIn
vit
ati
on
To
...
Pu
nk
bb
58d
847e7
efc1
62fe
e8cc
7f4
2ce
b8fc
We
Die
You
ng
Ali
ceIn
Ch
ain
sF
ace
lift
Alt
ern
ati
ved
1d
332e8
675375e9
7e6
c38f0
6601185c
Wor
ld’s
On
Her
oin
All
Mass
Ner
der
Pu
nk
4c1
ee1d
96d
416e7
0e6
d2e7
266199b
5a4
Cal
lin
gA
llS
kele
ton
sA
lkal
ine
Tri
oD
am
nes
iaP
un
k765e4
557a749e6
aed
ae5
6c5
a38845035
Qu
een
ofP
ain
Alk
alin
eT
rio
Alk
ali
ne
Tri
o,
Ho..
.P
un
ke4
a82e5
77d
bab
3a679845a689e1
e0f8
1
Cal
lin
gA
llS
kele
ton
sA
lkal
ine
Tri
oA
gony
An
dIr
ony
Pu
nk
97f1
91fb
9736eb
dd
ece8
c69081c1
b6ff
Up
Th
eC
uts
Aga
inst
Me!
New
Wav
eP
un
kf9
d7018
88e6
e49ee
a336b
062d
e8b
67f7
Bey
ond
the
Mir
age
Al
Di
Meo
la,
Paco
D..
.T
he
Gu
itar
Tri
oJazz
640b
35c
9afe
fc85d
b6aa47301d
dd
1f2
7
Cli
che
Gu
evar
aA
gain
stM
e!A
sT
he
Ete
rnal
Co..
.P
un
k0765d
6939b
ed976e3
4c6
2a979c1
af4
10
Ele
ctri
cE
ye
Ale
xS
kol
nic
kT
rio
Tra
nsf
orm
ati
on
Jazz
415d
59ff
d07d
75781342463798d
65b
a0
No
Rea
son
Aga
inst
All
Au
thori
tyD
estr
oyW
hat
Des
t...
Pu
nk
a04c4
669d
055ed
b2e6
8d
27d
83f6
b217e
San
sP
lan
ZA
gain
stT
omorr
ow’s
Sky
Th
eL
ost
Tap
esIn
die
f015cb
9cf
50682c5
d517187d
f6e6
eef4
Th
row
ing
Bro
ken
Sh
apes
Aga
inst
Tom
orr
ow’s
Sky
Ju
mp
Th
eH
edges
F..
.In
die
1f8
d9ed
3d
b93ac4
14d
5a9ae1
1b
9eb
e3a
Med
iocr
ity
Get
sY
ou..
.A
gain
stM
e!S
earc
hin
gfo
ra
F..
.P
un
k2eb
6631
6212d
41561642fa
66933a0ef
5
Th
eP
olit
ics
ofS
ta..
.A
gain
stM
e!R
einve
nti
ng
Axl
Rose
Pu
nk
768b
7ab
b853b
e6f8
3277e1
59707b
c934
An
chor
En
dA
Wil
hel
mS
crea
mM
ute
Pri
nt
Pu
nk
cef5
0cb
8ae0
00b
f79b
f4e7
2d
a357f1
e9
Ble
edB
lack
AF
IS
ing
Th
eS
orr
owP
un
k8373a879f3
7d
ab
a3ce
d662a1aae8
744f
Exsa
ngu
inat
ion
AF
IB
lack
Sail
sIn
Th
...
Pu
nk
7a66fc
b247d
64f4
66b
694d
6d
ca4d
454a
Ju
stifi
cati
onA
gain
stA
llA
uth
ori
tyA
llF
all
Dow
nP
un
k90d
e01e
d01733cd
b9c1
71597d
0c1
5439
110
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Th
eB
oyW
ho
Des
troy
...
AF
IA
llH
all
ow’s
E.P
.P
un
k03c3
560b
0a45c0
7f4
8cc
247ec
44f1
ec8
Yu
rfR
end
enm
ein
AF
IA
nsw
erT
hat
An
dS
...
Pu
nk
87e5
41a0608a46ee
05aa7f1
3fe
81e6
23
Cu
ltS
tatu
sA
FI
Ver
yP
rou
dO
fY
aP
un
k96d
baac4
a869cc
43d
1d
48d
f5912c4
006
Die
Wh
ile
We
reY
oun
gA
Wil
hel
mS
crea
mC
are
erS
uic
ide
Pu
nk
863e6
48b
77474c9
fad
80d
59376169b
a0
Had
My
Ch
ance
88F
inge
rsL
ou
ieB
ehin
dB
ars
Pu
nk
1fe
7c3
c289574d
d76605b
8357d
44d
0e4
Th
eS
oft
Sel
lA
Wil
hel
mS
crea
mR
uin
erP
un
k350297477c9
24a7f8
a738568742b
8248
Ju
dit
hA
Per
fect
Cir
cle
Mer
de
Nom
sM
etal
55e3
74e4
7f4
a7fb
2ca
5c6
4a41e4
3c9
fb
Bu
llet
pro
ofT
iger
AW
ilh
elm
Scr
eam
AW
ilh
elm
Scr
eam
Pu
nk
6e6
f81608261b
9c8
ef1d
4f4
6552910d
e
Str
aigh
tO
n7
Sec
ond
sT
he
Cre
wP
un
k7c6
fd7b
6e1
88354c5
7514fa
de3
e9108d
Hea
rtIn
Th
eH
and
O..
...
.An
dY
ouW
ill
Kn
o..
.S
ou
rce
Tags
&C
od
esIn
die
6d
e6b
235b
809ab
cd1ef
e4ec
44e4
3d
822
IW
asA
Tee
nag
eA
na.
..A
gain
stM
e!W
hit
eC
ross
esP
un
kd
fe06b
3f1
24f8
93d
83d
d82a0715001ef
New
Wav
eA
gain
stM
e!N
ewW
ave
Pu
nk
7b
21e4
bff
f344b
62d
602b
f2c9
d5d
ccc4
Eag
erA
gain
stT
omorr
ow’s
Sky
Ju
mp
Th
eH
edges
F..
.In
die
601fa
b833fc
588f3
f4544e3
143d
cab
14
La
Est
iba
Al
Di
Meo
la,
Paco
D..
.T
he
Gu
itar
Tri
oJazz
48d
0fd
f5fe
cbb
3f1
ad
fb8ee
86f8
45795
Lif
esty
leof
Reb
elli
onA
gain
stA
llA
uth
ori
tyD
estr
oyW
hat
Des
t...
Pu
nk
38857d
4b
ed8d
552e2
86744f0
2a4f5
eea
Lu
llab
yeL
eagu
eA
gain
stT
omorr
ow’s
Sky
Th
eL
ost
Tap
esIn
die
f098b
b40ae6
fcfa
c99233005b
3b
1ad
c6
T.S
.R.
Aga
inst
Me!
As
Th
eE
tern
al
Co..
.P
un
ka7f4
c5325c0
66d
8103328a076f4
f8636
Tra
nsf
orm
atio
nA
lex
Skol
nic
kT
rio
Tra
nsf
orm
ati
on
Jazz
c2756fe
99aacb
1546799aaff
4f0
01ed
6
Mia
mi
Aga
inst
Me!
Sea
rch
ing
for
aF
...
Pu
nk
9c8
c9f8
cc03c6
8ce
eda29991f3
3ae8
77
Pin
tsof
Gu
inn
ess
M..
.A
gain
stM
e!R
einve
nti
ng
Axl
Rose
Pu
nk
728c8
b8b
21f0
9130b
b39d
7d
6e3
483356
111
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
12:0
0A
MA
gain
stA
llA
uth
ori
tyA
llF
all
Dow
nP
un
kb
7a9ae6
922d
e955952f3
3312b
e840e7
c
Fam
ous
Fri
end
sA
nd
...
AW
ilh
elm
Scr
eam
Mu
teP
rint
Pu
nk
f77658c6
edd
12ad
bfc
bed
92ae2
fdd
fb6
Hal
f-em
pty
Bot
tle
AF
IA
nsw
erT
hat
An
dS
...
Pu
nk
16f5
4d
678c8
e4f2
e7973f1
3b
70c0
c2e1
Hal
low
een
AF
IA
llH
all
ow’s
E.P
.P
un
k20490af2
9384a6d
21988e0
356504fb
e5
Kil
lin
gIt
AW
ilh
elm
Scr
eam
Ru
iner
Pu
nk
134422285eb
d2d
cc8a9f0
26fb
886f8
cf
Por
phyri
aA
FI
Bla
ckS
ail
sIn
Th
...
Pu
nk
4fc
f656e5
7eb
8589d
b03a4b
77880a442
Th
eL
eavin
gS
ong
Pt.
2A
FI
Sin
gT
he
Sorr
owP
un
kca
0996b
052e9
f669a13f9
7b
929f9
4321
Ord
erA
vail
Liv
eat
the
Bott
o..
.P
un
ka2b
5a9f
08f0
0c0
5f1
21a7400b
797e8
89
Ou
trig
ht
Lie
s88
Fin
gers
Lou
ieB
ehin
dB
ars
Pu
nk
f24799443b
b1ec
da9e2
8b
9944fb
15662
Th
eH
orse
AW
ilh
elm
Scr
eam
Care
erS
uic
ide
Pu
nk
e3738417a76321a8900a5f3
3f7
59b
958
Wak
e-u
pC
all
AF
IV
ery
Pro
ud
Of
Ya
Pu
nk
80cd
f471b
83cf
199b
e5009d
85d
c88f2
e
Ros
eA
Per
fect
Cir
cle
Mer
de
Nom
sM
etal
fa66a9fb
a293f3
6992e7
8ed
7cf
3b
45d
6
Fu
nT
ime
AW
ilh
elm
Scr
eam
AW
ilh
elm
Scr
eam
Pu
nk
674d
19d
33b
5a84ca
c07c6
29b
4634b
167
Th
isIs
the
An
gry,
...
7S
econ
ds
Th
eC
rew
Pu
nk
d24f2
7c3e8
b8b
337d
bfe
7151b
723366a
How
Nea
r,H
owF
ar..
.An
dY
ouW
ill
Kn
o..
.S
ou
rce
Tags
&C
od
esIn
die
0193ae3
ced
25974b
99708ee
19599785c
Iron
Man
Ali
ceIn
Ch
ain
sD
irt
Alt
ern
ati
vec1
a51d
5f9
33ee
609f6
f6c9
082c2
b1acb
Wh
ite
Cro
sses
Aga
inst
Me!
Wh
ite
Cro
sses
Pu
nk
9d
be6
d1ff
570b
8d
3c5
8057175e0
e5273
All
Fal
lD
own
Aga
inst
All
Au
thori
tyA
llF
all
Dow
nP
un
k2a823a3b
7a0f1
9d
4845cd
a728c6
5016f
Fal
lC
hil
dre
nA
FI
All
Hall
ow’s
E.P
.P
un
k6cc
151e6
b2d
8967e9
8a6a29b
66b
256ca
He
Wh
oL
augh
sL
ast
AF
IV
ery
Pro
ud
Of
Ya
Pu
nk
44e5
bcd
b65a89026a9d
160300b
386b
5a
112
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Mis
eria
Can
tare
–Th
...
AF
IS
ing
Th
eS
orr
owP
un
k755446b
4d
499523d
9ee
41e8
ef6090ac4
Mu
teP
rint
AW
ilh
elm
Scr
eam
Mu
teP
rint
Pu
nk
76b
af8
7b
c8eb
c4ee
29af7
4d
a28f5
a019
Str
engt
hT
hro
ugh
Wo.
..A
FI
Bla
ckS
ail
sIn
Th
...
Pu
nk
98c0
8b
644b
2259d
ad
8ad
3afc
74b
99663
Th
eK
ing
isD
ead
AW
ilh
elm
Scr
eam
Ru
iner
Pu
nk
4a0b
5d
f76f9
865d
8d
c435b
a86150a69d
Tw
oO
fA
Kin
dA
FI
An
swer
Th
at
An
dS
...
Pu
nk
911cd
bc3
4949d
e445a1172016e3
58fb
4
5T
o9
AW
ilh
elm
Scr
eam
Care
erS
uic
ide
Pu
nk
c3f2
d10
933e6
873ea
80d
d5ef
54089237
Exp
lan
atio
n88
Fin
gers
Lou
ieB
ehin
dB
ars
Pu
nk
79b
ac1
7176f3
5c6
0b
30a687cd
9512f0
e
Mag
dal
ena
AP
erfe
ctC
ircl
eM
erd
eN
om
sM
etal
08061f7
d017d
ac9
a95786f5
760a49a79
Eve
ryG
reat
Sto
ryH
...
AW
ilh
elm
Scr
eam
AW
ilh
elm
Scr
eam
Pu
nk
0296c2
052c0
bf9
784d
933d
61ab
e68f0
e
Bau
del
aire
...A
nd
You
Wil
lK
no..
.S
ou
rce
Tags
&C
od
esIn
die
a3735c4
4f9
edb
69993d
7921961a93a70
Lif
eO
nT
he
Roa
dA
llM
ass
Ner
der
Pu
nk
0ef
1b
0174d
aa8ad
3e5
a6c1
078f7
20ad
5
Not
Ju
stB
oys
Fu
n7
Sec
ond
sT
he
Cre
wP
un
k77ef
59d
a82a5fc
d49191a469ab
1874d
3
Rep
arat
ion
88F
inge
rsL
ou
ie88
Fin
ger
sL
ou
ie..
.P
un
k920b
a30
f32fe
71e8
6fa
86b
2375f0
cb65
Har
k!
the
Her
ald
An
...
Bin
gC
rosb
yF
rank
...
Ch
rist
mas
Jazz
96e1
4345b
837229b
f9e4
8f2
21eb
dd
439
30S
ec.
Son
gA
gain
stA
llA
uth
ori
tyD
estr
oyW
hat
Des
t...
Pu
nk
d932fc
589448e4
11096cc
94b
9843b
3a5
Pen
tU
p88
Fin
gers
Lou
ieB
ehin
dB
ars
Pu
nk
f956b
4e3144b
241d
20603e9
a8c2
f6574
Th
eH
ollo
wA
Per
fect
Cir
cle
Mer
de
Nom
sM
etal
139ac4
259aafc
07c6
d8a6a435d
9167e8
Au
stra
lias
AW
ilh
elm
Scr
eam
AW
ilh
elm
Scr
eam
Pu
nk
2f8
787c1
efae9
5f6
675d
cbc7
7178b
324
An
oth
erM
orn
ing
Sto
ner
...A
nd
You
Wil
lK
no..
.S
ou
rce
Tags
&C
od
esIn
die
b32aaf6
384d
ceb
181d
2ee
851acd
fa86d
Defi
nit
eC
hoi
ce7
Sec
ond
sT
he
Cre
wP
un
kb
4e1
23a
85d
d40f1
d2056b
51a5921b
496
113
Tab
leB
.1–FullEvaluationDatasetContinued
Song
Artist
Album
Genre
MD5
Slo
wC
hor
us
Ove
rlap
88F
inge
rsL
ou
ie88
Fin
ger
sL
ou
ie..
.P
un
kf2
cbc4
bf2
936c7
31075f7
a9cf
8b
7f3
18
Car
eer
Su
icid
eA
Wil
hel
mS
crea
mC
are
erS
uic
ide
Pu
nk
f3a3e0
9897d
fb159073d
f1d
5d
4b
a9e2
4
Zer
oF
risc
oB
raid
Fra
nkie
Wel
fare
B..
.In
die
690a79d
fcef
0372f9
43d
c4b
013009868
Her
e’s
You
rW
arn
ing
7S
econ
ds
Th
eC
rew
Pu
nk
24f9
c12d
fb84f8
b94718f4
71120d
730c
IW
ipe
My
A**
Wit
h..
.A
Wil
hel
mS
crea
mC
are
erS
uic
ide
Pu
nk
51ee
4f0
043069e3
f41c3
cb52b
c6b
1200
ItW
asT
her
eT
hat
I...
...A
nd
You
Wil
lK
no..
.S
ou
rce
Tags
&C
od
esIn
die
27707e7
d00b
d2036e8
548c2
698f5
11b
7
Ou
tT
her
e88
Fin
gers
Lou
ie88
Fin
ger
sL
ou
ie..
.P
un
k505626e5
0b
828fc
415e7
a809846c3
a88
Fil
e13
AF
IV
ery
Pro
ud
Of
Ya
Pu
nk
3262b
f7036fe
a610ca
b9d
c2aed
5af8
1b
Hom
age
...A
nd
You
Wil
lK
no..
.S
ou
rce
Tags
&C
od
esIn
die
4b
a200e
93b
3ed
7b
2a690aa81b
8fe
e497
Cle
nch
edF
ists
,B
la..
.7
Sec
ond
sT
he
Cre
wP
un
k91e5
14b
8af4
59fd
7e1
e05f0
51f8
37a6a
Nar
rati
veO
fS
oul
A..
.A
FI
Bla
ckS
ail
sIn
Th
...
Pu
nk
a7b
0790
c1e0
5012d
aa21c6
feca
b12142
Th
eP
ool
AW
ilh
elm
Scr
eam
Ru
iner
Pu
nk
86eff
07e2
414963c4
356d
826500b
772c
114
References
[1] Gul Abdulnabi Agha. Actors: a model of concurrent computation in distributedsystems. MIT Press, Cambridge, MA, 1985.
[2] Akka. http://akka.io/. Last visited 11/18/2013.
[3] Krste Asanovic, Ras Bodik, Bryan Christopher Catanzaro, Joseph James Gebis,Parry Husbands, Kurt Keutzer, David A. Patterson, William Lester Plishker,John Shalf, Samuel Webb Williams, and Katherine A. Yelick. The land-scape of parallel computing research: A view from berkeley. Technical ReportUCB/EECS-2006-183, EECS Department, University of California, Berkeley,Dec 2006.
[4] Sebastian Bock, Florian Krebs, and Markus Schedl. Evaluating the online capa-bilities of onset detection methods. In ISMIR, pages 49–54, 2012.
[5] Stuart Bray and George Tzanetakis. Distributed audio feature extraction formusic. In ISMIR, pages 434–437, 2005.
[6] Judith C Brown and Miller S Puckette. An efficient algorithm for the calculationof a constant q transform. The Journal of the Acoustical Society of America,92:2698, 1992.
[7] Jamie Bullock and UCEB Conservatoire. Libxtract: A lightweight library foraudio feature extraction. In Proceedings of the International Computer MusicConference, volume 43, 2007.
[8] Chris Cannam, Michael O. Jewell, Christophe Rhodes, Mark Sandler, and Markd’Inverno. Linked data and you: Bringing music research software into thesemantic web. Journal of New Music Research, 39(4):313–325, 2010.
[9] Akka Clustering. http://doc.akka.io/docs/akka/2.2.1/common/cluster.
html. Last visited 02/10/2014.
[10] John Cocke. Global common subexpression elimination. ACM Sigplan Notices,5(7):20–24, 1970.
[11] BBC Vamp Plugin Collection. http://github.com/bbcrd/bbc-vamp-plugins.Last visited 02/29/2014.
115
[12] Steven Davis and Paul Mermelstein. Comparison of parametric representationsfor monosyllabic word recognition in continuously spoken sentences. Acoustics,Speech and Signal Processing, IEEE Transactions on, 28(4):357–366, 1980.
[13] Jeffrey Dean and Sanjay Ghemawat. Mapreduce: simplified data processing onlarge clusters. Communications of the ACM, 51(1):107–113, 2008.
[14] Giuseppe DeCandia, Deniz Hastorun, Madan Jampani, Gunavardhan Kakula-pati, Avinash Lakshman, Alex Pilchin, Swaminathan Sivasubramanian, PeterVosshall, and Werner Vogels. Dynamo: amazon’s highly available key-valuestore. In SOSP, volume 7, pages 205–220, 2007.
[15] Francois Deliege, Bee Yong Chua, and Torben Bach Pedersen. High-level audiofeatures: Distributed extraction and similarity search. In ISMIR, pages 565–570,2008.
[16] Simon Dixon. Onset detection revisited. In Proceedings of the 9th InternationalConference on Digital Audio Effects, volume 120, pages 133–137, 2006.
[17] Florian Eyben, Martin Wollmer, and Bjorn Schuller. Opensmile: the munichversatile and fast open-source audio feature extractor. In Proceedings of theinternational conference on Multimedia, pages 1459–1462. ACM, 2010.
[18] Roy T Fielding and Richard N Taylor. Principled design of the modern webarchitecture. ACM Transactions on Internet Technology (TOIT), 2(2):115–150,2002.
[19] Vamp Plugin Framework. http://www.vamp-plugins.org/. Last visited03/21/2014.
[20] Matteo Frigo and Steven G. Johnson. The design and implementation of FFTW3.Proceedings of the IEEE, 93(2):216–231, 2005. Special issue on “Program Gen-eration, Optimization, and Platform Adaptation”.
[21] Olivier Gillet and Gael Richard. Automatic transcription of drum loops. InAcoustics, Speech, and Signal Processing, 2004. Proceedings.(ICASSP’04). IEEEInternational Conference on, volume 4, pages iv–269. IEEE, 2004.
[22] Neil J. Gunther. Guerrilla Capacity Planning. Springer Berlin Heidelberg, 2007.
[23] Naohiro Hayashibara, Xavier Defago, Rami Yared, and Takuya Katayama. Theϕ accrual failure detector. In Reliable Distributed Systems, 2004. Proceedings ofthe 23rd IEEE International Symposium on, pages 66–78. IEEE, 2004.
[24] Gregor Hohpe and Bobby Woolf. Enterprise integration patterns: Designing,building, and deploying messaging solutions. Addison-Wesley Professional, 2004.
[25] Scala Programming Language. http://www.scala-lang.org/. Last visited11/18/2013.
116
[26] Olivier Lartillot, Petri Toiviainen, and Tuomas Eerola. A matlab toolbox formusic information retrieval. In Data analysis, machine learning and applications,pages 261–268. Springer, 2008.
[27] Benoit Mathieu, Slim Essid, Thomas Fillon, Jacques Prado, and Gael Richard.Yaafe, an easy to use and efficient audio feature extraction software. In ISMIR,pages 441–446, 2010.
[28] Daniel McEnnis, Cory McKay, and Ichiro Fujinaga. Overview of omen. In ISMIR,pages 7–12, 2006.
[29] Daniel McEnnis, Cory McKay, Ichiro Fujinaga, and P Depalle. jaudio: Additionsand improvements. In ISMIR, pages 385–386, 2006.
[30] Cory McKay, Ichiro Fujinaga, and Philippe Depalle. jaudio: A feature extractionlibrary. In Proceedings of the International Conference on Music InformationRetrieval, pages 600–3, 2005.
[31] University of Colorado Colorado Springs. EAS Cloud. http://www.uccs.edu/
eas/eas_cloud.html. Last visited 03/15/2014.
[32] QM Vamp Plugins. http://isophonics.net/QMVampPlugins. Last visited02/29/2014.
[33] Apple Press. http://www.apple.com/pr/library/2013/02/
06iTunes-Store-Sets-New-Record-with-25-Billion-Songs-Sold.html.Last visited 03/29/2014.
[34] Apache Software Foundation. HDFS Short-Circuit Local Reads. http:
//hadoop.apache.org/docs/r2.2.0/hadoop-project-dist/hadoop-hdfs/
ShortCircuitLocalReads.html. Last visited 03/05/2014.
[35] Devon Bryant. Scalable Audio Feature Extraction. Source Repository. http:
//github.com/devonbryant/safe. Last visited 04/10/2014.
[36] Devon Bryant. Numerical Computing With Scala. http://devonbryant.
github.io/blog/2013/03/03/numerical-computing-with-scala/. Last vis-ited 03/25/2014.
[37] Devon Bryant. MIR Tool Evaluation Scripts. http://github.com/
devonbryant/mir-eval-scripts. Last visited 04/10/2014.
[38] Konstantin Shvachko, Hairong Kuang, Sanjay Radia, and Robert Chansler.The hadoop distributed file system. In Mass Storage Systems and Technologies(MSST), 2010 IEEE 26th Symposium on, pages 1–10. IEEE, 2010.
[39] Spotify. http://press.spotify.com/us/information/. Last visited03/29/2014.
117
[40] Herb Sutter and James Larus. Software and the concurrency revolution. Queue,3(7):54–62, September 2005.
[41] George Tzanetakis and Perry Cook. Marsyas: A framework for audio analysis.Organised sound, 4(3):169–175, 2000.
[42] Robbert Van Renesse, Yaron Minsky, and Mark Hayden. A gossip-style failuredetection service. In Middleware’98, pages 55–70. Springer, 1998.
[43] Eclipse Public License version 1.0. http://www.eclipse.org/org/documents/epl-v10.html. Last visited 04/10/2014.
118