Chapter 12

Room Acoustics II:The Listener and the Room

Acoustic Extremes Anechoic Chamber

rock-wool wedges to prevent echoes from walls, ceilings, and floors.

No one allowed in room and no furniture either to prevent scattering.

Even better is to deliver the sound electronically through headphones (acoustically sterile)

Anechoic Chamber

Real Room Human ears are much better in

discriminating small changes in pitch, loudness, and tone color in a real room.

Trumpet Experiment

Trumpeter plays a steady A3 (220 Hz). Harmonics at 440, 660, 880, 1100,

1320, etc Below 1200 HZ the source is small

compared to the wavelength

Source size

Human Hearing Response

The human nervous system makes a running average amplitude (loudness) of the partials based on information received from the two ears.

Useful for partials below 1000 Hz. Pressure fluctuations can be correlated

over one-half wavelength. Over distances more than half wavelength

there is no correlation

Correlation length At 1000 Hz the wavelength is 34.5 cm.

Your two ears are more than ½-wavelength apart and they pick up independent views of the room.

At the trumpeter’s 220 Hz (wavelength = 1.57 m) the ears are less than ½-wavelength apart and the sounds are well correlated.

They are too close together to be useful in getting extra information by the averaging process.

Head Movement If you move your head a little with

the music, you pick up enough variations in the relative partial amplitudes to improve the averaging process.

Low frequency information is limited again because the distances are still small compared to the wavelength and the information at the two ears is essentially the same.

Swaying of the musician has the same effect.

Short Period/Higher Frequency

Our nervous systems can also accumulate the averages it is forming over short periods of time, in order to take advantage of moving objects in the room

Small motions can be exploited for high frequencies only (above 500 Hz) - this is because of the size of the body in comparison to the wavelengths of the sounds.

Attack and Decay Comprised of two parts - what the

instrument is doing and what the room is doing. We are interested in the room here.

Low Frequency Attack and Decay Source acts as a point and sound

emanates in all directions with equal strength (homogeneously).

We get the direct sound and a few milliseconds later, six reflections from the walls, etc. The reflections off of large, flat surfaces do

not alter the waves. The decay is the attack in reverse.

High Frequency Attack and Decay Source is now directional (the

direction the bell is pointed). The direct sound may have much

higher amplitude now than the reflections, if the bell is pointed at us.

Someone away from the line-of-sight gets a different mix of amplitudes.

Joseph Henry First secretary of the Smithsonian

Acoustics Applied to Public Buildings He wanted to measure the shortest

time in which a reflected sound would be heard as distinct from the direct sound. At greater times echoes would be

distracting

Henry’s Findings If the echo traveled an extra 20 m,

it can be heard as distinct. 20 m at 345 m/s is a time delay of 58

ms - call it 60 ms Look at sound arriving well within

that time period – say 35 ms.

Precedence Effect Consider two clicks delivered to two

speakers separated by a few feet. Track the time delay t between clicks.

I will describe the effect first and then there is a sound file for you to hear it. The sound file has clicks delivered to the

speakers and t is varied.

Precedence Effect When t is between about 5 and 20 ms,

sound seems to emanate from the speaker emitting the leading click.

When t is very short, there is complete fusion, and a "phantom" image occurs between the two loudspeakers

When t is very long, fusion no longer occurs and each click is perceived as a separate sound source.

Listen

Clifton Effect Several click pairs (12 ms

separation) left speaker first Reverse the order of the speaker

clicks (right speaker first) At first the two clicks seem to be

separated They then merge together and appears

to come from the lead speakerTry it!

Summary The ear will combine a set of

reduplicated sounds (echoes) and hear them as one provided… that they arrive within about 35 ms of

each other, and that the waveforms are sufficiently similar.

The one tone is heard without any delay.

Summary – cont’d The perceived time of arrival is that of

the first sound. The loudness may be greater than the

first sound alone. The apparent position of the sound is the

position of the first sound. The effect is present even when the later

arrivals have more amplitude. but less than about three times the

amplitude of the first

Design of a Speaker System

Imagine a church with a long nave, so long that the preacher’s voice is not loud enough to carry to the back. The speaker's mouth acts as a point source and

the sound spreads out uniformly from there. The amplitude will be very small in the back.

Some of the echoes will arrive after the 35 ms cutoff for the Precedence Effect to work. These echoes then become annoying distractions.

Handling the Front Place a speaker above and behind

the preacher's head. Project sound down the length of the

hall. Speaker's output has only a slight

delay compared to the direct sound. Precedence Effect will make the sound

seem to originate from the voice.

Handling the Back Place a few non-directional speakers

toward the back Electronic delay so that the sound from the

front speaker arrives slightly before the sound from the back speakers. Precedence effect we hear the sound arriving

from the front. Back speakers cannot deliver more than three

times the original amplitude Non-directional so as not to announce their

position.

Auditory System The ear/brain has the ability to

focus on particular sounds in a room filled with sound

Easier in a room than outside, since the room provides reflections and scattering to help fill in the information.

Head Experiment Set two microphones separated by

the size of your head. We take measurement with and

without the head present.

Without Head Source slightly closer to right.

Left

Right

With Head Shadowed (left) ear has less amplitude Right has much higher signal The two are quite different Clearly the listener’s head has an influence on the sound

Left

Right

Position CuesA

B

L

M

Listener Clues Here only one ear is used to simplify We start with the path lengths equal

(MAL = MBL) If either listener or musician moves,

then the ear detects differences in the amplitude of the paths and can detect the change in position.

Additional information With two ears there is more

information to help in locating sound

Caution: we are now considering distant listeners If the listener is close (< 1 wavelength)

then scattering is different

Aural Perception A person who can move his head

processes the information better than one who cannot

Binaural hearing is better than monaural hearing

Headphone disadvantage we are then deprived of the cross-correlated

clues coming from the room Used in perception studies to limit sounds

More Aural Perception We can take advantage of distinctive features

in a sound pattern to recognize the sound We learn the scattering pattern of nearby

objects very quickly and use these to help distinguish the effect of the objects from the original sound

We can take advantage of several types of auditory information simultaneously

We can detect short time period changes in the sound source.

Aural Processing On first arrival of a sound, we

make a quick preliminary judgment as to the position and nature of the source.

We then use the precedence effect to fill in information in the first 35 ms.

Aural Processing (cont’d) Other processors are at work to

help us sort frequency and time of arrival information. There is evidence that we can sometimes distinguish sound separated by 30 s.

Sounds separated by more than 60 ms are heard as distinct.

Flutter Echoes Clap your hands in a large room

Rapidly repeating series of echoes Period equal to the round trip travel

time between walls or floor and ceiling Usual frequency is several a second, it

may sound like something fluttering Frequency may even be high enough

to assign a pitch

Perception of Repeated Notes Trumpet player plays a quick series

of notes (2 – 9/sec) Listener can hear each note Oscilloscope gives good agreement at

low repetition rate Irregular jumble at high repetition rate

Irregularities come from the room The ear can deal with these

Loudspeaker Response

ResonantFrequency

High Frequency

Cut-off

Speaker Response Regions At frequencies below speaker

resonance, response falls rapidly This is the frequency that the cone

oscillates at if displaced from rest Mid-range is approximately constant High frequency cut-off

Cone is larger than a few wavelengths of the sound

Rule of Thumb If wavelength of sound is shorter

than half the circumference of the speaker cone, then the response is poor

Ex. Consider a 12-inch diameter speaker C = 2r = (2)(3.1416)(6 in.) = 37.7 in. = ½ C = 18.85 in. = 1.57 ft. f = v/ = (1133 ft/s)/(1.57 ft.) = 721 Hz

High Frequency Cut Off In our example frequencies above 700

Hz are not well reproduced. At 1400 Hz the response is ¼th the 700 Hz

response At 2800 Hz it is 1/16th as much as at 700

Hz The beam pattern is less homogeneous

above the high frequency cut off

12-inch speaker at 250 Hz

12-inch speaker at 1000 Hz

12-inch speaker at 4000 Hz

Typical Speaker Arrangement

Woofer

Tweeter

Mid-range

Lows

Middles

Highs

From Amplifier

D

Problems at Crossover Frequency Consider frequencies near where

one speaker hands off to another We can have a situation of two

sources of almost equal strength Speaker separations by D = ½ , 3/2

, 5/2 , etc. will lead to total destructive interference for most room modes

Crossover Example For a frequency of 2000 Hz

(crossover between mid-range and tweeter)

Wavelength of the sound is… = v/f = (345 m/s)/(2000 Hz) = 0.17

m = 17 cm The affected spacings would be 8.6

cm, 25.9 cm, 43.1 cm, etc.

Other Problems of Crossover Electrical circuits controlling the

crossovers assume that the electrical responses of the speakers do not change with frequency. But they usually do, resulting in

irregular behavior far from the crossover frequencies

Getting Too Fancy in Testing Problems can be overlooked if

speaker tests are performed in anechoic chambers. The aim of these rooms is to record the

sound from the source before the room modes have a chance to affect it.

Multiple speaker problems will not show up.

Cheaper Speaker Systems Sometimes you can adjust the

cone shape to give ok response over a wider frequency range only one source, no multi-speaker

cancellations exist no crossover electrical circuit is

required, no electrical problems exist

Other Solutions Two speaker system without electrical

circuits Woofer will work best on the lows, tweeter on

the highs A simple circuit allows more power to the

tweeter at higher frequency In more sophisticated versions, the tweeter and

woofer are about ¼ out of phase at the crossover frequencies, avoiding the cancellation

Your hearing is good at rejecting unwanted sounds – so a lot of this is overkill.

Impulsive Sound Generator Clapper below works well for high

frequencies Clap hands or bang bucket for low

frequencies

Experiments on Flat Wall Find the shortest distance you can

stand to the wall and detect distinct echoes Measure the distance to the wall Recall the echo has to make the round

trip Calculate the time required for echo Any shorter distance and the Precedence

Effect takes over

Experiment 2 Use the wall to estimate distance Your footstep echoes can provide

information on the distance to the wall while you walk toward it. Even though the echoes will come

within the 35 ms time domain of the precedence effect, your brain can still process the information.

Experiment 3 Drive down a quiet street with only

the right front (and then the right rear) windows open. Pay attention to the way scattered

sound comes to you from curbs, trees, etc.

Experiment 4 John Henry found he could understand

a speaker outdoors 100 ft. in front, but not when 30 ft. behind. Compare male speakers to female.

Frequency(Hz)

Amplitude Ratio (forward/backward)

100 1

200 1.5

1000 3

5000 8

Verify the Precedence Effect with a Home Stereo Set it for monaural output (same

signal to both speakers) Use the balance control to adjust

the gain of each speaker You should see that the nearby

speaker seems to be the source, even if the farther speaker is stronger