INTEGRATION OF VOICE SERVICES IN INTERNET APPLICATIONS By Eduardo Carrillo (lecturer), J. J Samper,...

Post on 28-Mar-2015

216 views 0 download

Tags:

Transcript of INTEGRATION OF VOICE SERVICES IN INTERNET APPLICATIONS By Eduardo Carrillo (lecturer), J. J Samper,...

INTEGRATION OF VOICE SERVICES

IN INTERNET APPLICATIONS

By

Eduardo Carrillo (lecturer),

J. J Samper, J.J. Martínez-Durá

Universidad Autónoma de Bucaramanga Universidad de Valencia

Colombia-Spain

Agenda

• Overview

• Developed Trials (our prototype)

• Arquitecture of Voice Internet Application

• Server Side Technologies and XML

• VoiceXML

• VoiceXML and our tests

• Conclusions

Overview

Main concepts related to accessing the internet contents by speech

via a telephone

OverView (cont)

Financial,Banking

ParkingPharmaciesHospitals

WeatherEntertainment

Reservation,Billing,Others

Reservation,Billing,Others

Overview (cont)

Developed trials

Our prototype

presents

weather information

of some european cities

The <<dynamic>> information is available via a normal telephone

Phone number

(34) 90188412

ID: 61 21 21 21

Pin: 1234

Developed Trials (cont.)

• Web Server:Apache

•Server Side Scripting Technology: PHP/Xalan

•Voice Markup Language: VoiceXML

Developed Trials (cont.)

Developed trials (cont.)

Arquitecture of Voice Internet Application

Arquitecture of Voice Internet Application (cont.)

Server Side Technologiesand XML

Server Side Technologies and XML (cont.)

Server Side Technologies andXML (cont.)

“VoiceXML is designed for creating audio dialogs

that feature synthesized speech, digitized audio,

recognition of spoken and DTMF key input,recording of spoken input,

telephony, and mixed-initiative conversations”

W3C Consortium(Voice Extensible Markup Language,

VoiceXML Version 2.0)

“VoiceXML is designed for creating audio dialogs

that feature synthesized speech, digitized audio,

recognition of spoken and DTMF key input,recording of spoken input,

telephony, and mixed-initiative conversations”

W3C Consortium(Voice Extensible Markup Language,

VoiceXML Version 2.0)

VoiceXML

VoiceXML is only used

to describe the conversationbetween

the voice portaland the caller

(the user interfaceof the voice application)

VoiceXML is only used

to describe the conversationbetween

the voice portaland the caller

(the user interfaceof the voice application)

VoiceXML (cont.)

VoiceXML and our tests

Our prototype

presents

weather information

of some european cities

The <<dynamic>> information is available via a normal telephone

Phone number

(34) 90188412

VoiceXML and our tests (cont.)

<?xml version="1.0" ?>

<!DOCTYPE vxml PUBLIC '-//Voxpilot/DTD VoiceXML 1.0//EN' 'http://dtd.voxpilot.com/voice/1.0/voxpilot_voicexml-1.0.dtd'>

<vxml version="1.0" >

<form>

<block>Welcome to the European meteorological information service.

<goto next="meteo.vxml" />

< - - This is a comment : goto is the hyperlink to the next document - - >

</block>

</form>

</vxml>

VoiceXML and our tests (cont.)

Meteo.vxml

<?xml version="1.0"?>

<!DOCTYPE vxml PUBLIC '-//Voxpilot/DTD VoiceXML 1.0//EN' 'http://dtd.voxpilot.com/voice/1.0/voxpilot_voicexml-1.0.dtd'>

VoiceXML and our tests (cont.)<form>

<field name="city">

<prompt>

Please select the city: Paris, Rome, London or Madrid.

</prompt>

<grammar>

[paris rome london madrid]

</grammar>

<noinput>please choose one of the following cities: Paris, Rome, London or Madrid</noinput>

<nomatch>please say Paris, Rome, London or Madrid</nomatch>

</field>

VoiceXML and our tests (cont.)

<filled>

<submit namelist="city" next="meteo.php" method="get" enctype="application/x-www-form-urlencoded" />

</filled>

</form>

</vxml>

Conclusions

ABC #$%

In this article a process to carry out a voice interaction with an application that generates dynamic content based on VoiceXML, has been described. With this structure, it is possible to access the server contents from a telephone, both fixed and mobile

Conclusions (cont.)

ABC

ABCABC

The voice portals inherit the advantages and disadvantages of technologies like the voice synthesis and voice recognition.

The quality of the recognition, the background noise and the difficulty to maintainance of user's attention of must be considered with special attention.

Conclusions (cont.)

ABC

ABCABC

Also in the process of text-to-speech conversion it is important to simulate human-to-human conversation as best as possible.

Conclusions (cont.)

xxxxxx

ABC

It has been possible to identify different lines of research such as the development of platforms for the detection of different accents, multilingual applications, natural language processing integrated in voice portals and the development of systems to integrates voice speech with other devices in multimodal applications.