date: Fri, 16 Dec 2005 14:38:17 -0600 content-transfer-encoding: quoted-printable content-location: http://www.colligo.us/Products/rtsd.htm x-mimeole: Produced By Microsoft MimeOLE V6.00.2900.2180 from: subject: =?Windows-1252?Q?AT&T_Natural_Voices=E4?= content-type: text/html; charset="Windows-1252" mime-version: 1.0 AT&T Natural Voic= es

AT&= T=20 Natural Voices=E4= =20

Text-To= -Speech=20 Engines

System Developer=92s Guide

Server, Server-Lite, and Desktop Editions

US English, Latin Am= erican=20 Spanish, UK English, German, and French

Release 1.4

&n= bsp;

&= nbsp;

=D3= 2001, 2002 by AT&T, All rights=20 reserved. Natural Voices is= a=20 registered trademark of AT&T Corporation. IBM is a registered trademark of Internationa= l=20 Business Machines Corporation. =20 Microsoft is a registered trademark and Windows, Visual Basic, and = Visual=20 C++ are trademarks of Microsoft Corporation. Pentium is a trademark of Intel= =20 Corporation.=20 Windows NT is a registered trademark of Microsoft Corporation. Sun= =20 Microsystems Solaris is a trademark of Sun Microsystems, Inc. Sparc is a=20 registered trademark of Sparc International, Inc. Intel is a registered=20 trademark of Intel Corporation.

Printed in the United States o= f=20 America.

10 9 8 7 6 5 4 3 2 1 =20

Contents

About this Guide........................................= ..........................................................................= ..........................................=20 vii

1 =20 Introduction. 7

1.1. AT= &T=20 Natural Voices Text-To-Speech. 7

1.2 Th= e=20 Text-to-Speech Synthesis Problem..

1.3 Re= lease=20 1.4 Features.

1.4 Sy= stem=20 Components and Features.

1.5 Su= pported=20 Languages.

1.6 Th= is=20 Guide.

1.7 Ot= her=20 Sources of Information.

2 =20 Installation.

2.1 In= stalling=20 the AT&T Natural Voices TTS Software.

2.2 Di= sk=20 Requirements.

2.3 Me= mory=20 Requirements.

2.4 In= stalling=20 AT&T Natural Voices TTS Engines on Microsoft Windows.

2.5 In= stalling=20 the AT&T Natural Voices TTS Software on UNIX Machines.

2.5.1 =20 Installing the AT&T Natural Voices Server-Lite and Deskt= op=20 Editions.

2.5.2 =20 Installing the Server Edition TTS Server on UNIX Servers.

2.5.3 =20 Installing the Server Edition TTS Client on UNIX Machines.

3 Using th= e TTS=20 Server Engines.

3.1 Se= rver=20 Command Line Arguments.

3.2 Ru= nning=20 the TTS Server as a Microsoft Service.

3.2.1 =20 Installing the ATT_TTSService Service.

3.2.2 =20 Running the ATT_TTSService service.

3.2.3 =20 Uninstalling the ATT_TTSService.

3.2.4 =20 Modifying the ATT_TTSService Parameters.

3.3 Su= pporting=20 Multiple Voices and Languages.

3.4 Ru= nning=20 the Client

3.4.1 =20 Microsoft Windows SAPI 4.0 TTSApp.

3.4.2 =20 Microsoft Windows SAPI 5.1 TTSApp.

3.4.3 =20 Command Line TTS Clients.

3.5 Se= rver=20 Output and Error Messages.

3.5.1 =20 Initialization Messages.

3.5.2 =20 Shutdown Messages.

3.5.3 =20 Voice Inventory Messages.

3.5.4 =20 Client Messages.

3.6 SN= MP.

3.6.1 =20 Configuring SNMP on Windows NT/2000/XP.

3.6.2 =20 Verifying that the Microsoft SNMP Service is installed.

3.6.3 =20 Configuring SNMP on Unix systems.

3.6.4 =20 Configuring the snmpd.conf configuration file.

3.6.5 TTS=20 MIB files.

3.7 Us= ing the=20 Server Edition with Nuance.

4 Running = the=20 Server-Lite and Desktop Edition.

4.1 Mi= crosoft=20 Windows SAPI 4.0 TTSApp.

4.2 Mi= crosoft=20 Windows SAPI 5.1 TTSApp.

4.3 Co= mmand=20 Line TTS Applications.

5 Client C= ++=20 SDK..

5.1 Ge= tting=20 Started.

5.1.1 SDK=20 Headers.

5.1.2 =20 Compiler Definitions.

5.1.3 SDK=20 Libraries.

5.2 =20 Programming Conventions.

5.2.1 =20 Return Codes.

5.2.2 =20 CTTSRefCntObject

5.2.3 =20 UTF-8 Strings.

5.2.4 C++=20 Standard Template Library.

5.3 Cr= eating=20 An Engine.

5.3.1 =20 Build A Configuration Structure.

5.3.2 =20 Creating an Engine.

5.3.3 =20 Creating an Engine for the Server Edition.

5.3.4 =20 Creating an Engine for the Server-Lite and Desktop Editions<= SPAN=20 style=3D"DISPLAY: none; COLOR: windowtext; TEXT-DECORATION: none; text-und= erline: none; mso-hide: screen">.

5.4 Re= ceiving=20 Engine Notifications And Messages.

5.4.1 =20 CTTSSink Notifications.

5.4.2 =20 CTTSNotification Object

5.4.3 =20 TTSWordNotification structure.

5.4.4 =20 TTSPhonemeNotification structure.

5.4.5 =20 CTTSErrorInfoObject

5.5 =20 Initializing And Shutting Down The Engine.

5.6 Se= tting=20 The Voice.

5.6.1 =20 TTSVoice Structure.

5.6.2 =20 Enumerating The Voices.

5.6.3 =20 Setting The Voice.

5.7 Se= tting=20 The Audio Format

5.7.1 =20 TTSAudioFormat Structure.

5.7.2 =20 Enumerating The Audio Formats.

5.7.3 =20 Setting the Audio Format

5.8 Sp= eaking=20 Text

5.8.1 =20 Speaking Text

5.8.2 =20 Speaking CTTSTextFragment objects.

5.9 Se= tting=20 Volume and Rate.

5.10 Stopping= The=20 Engine.

5.11 Retrievi= ng=20 Phonetic Transcriptions.

5.12 Custom=20 Dictionaries.

5.12.1 Cr= eating a=20 Dictionary.

5.12.2 Up= dating a=20 Dictionary.

5.12.3 Cl= earing=20 and Deleting a Dictionary.

5.12.4 Ch= anging=20 Search Order

5.13 Optional= =20 Operating System Specific Classes.

5.13.1 =20 CTTSWin32AudioPlayer

5.13.2 =20 CTTSWin32AudioWriter

5.13.3 =20 CTTSUnixAudioPlayer

5.13.4 =20 CTTSUnixAudioWriter

5.14 Changes = from=20 AT&T Natural Voices 1.0 Release.

5.14.1 Ne= w Engine=20 Models.

5.14.2 Li= brary=20 names have changed.

6 Java Spe= ech API=20 Implementation.

6.1 =20 Requirements.

6.2 =20 Installation.

6.3 Co= mpiling=20 the Examples.

6.4 Ru= nning=20 the Examples.

6.5 Us= ing=20 AT&T Natural Voices JSAPI Implementation.

6.6 Av= ailable=20 Synthesizer Voices.

6.7 Av= ailable=20 Synthesizer Modes.

6.8 =20 Differences Between the JSAPI Specification and the AT&T= =20 Implementation.

6.8.1 =20 Audio.

6.8.2 =20 Vocabulary.

6.8.3 =20 Queuing.

6.8.4 =20 XML.

6.8.5 =20 Voices.

6.8.6 =20 Synthesizer

6.8.7 =20 Engine.

6.8.8 =20 Permissio= ns.

7 SAPI 4 T= ext=20 Markup.

7.1 Co= m..

7.2 Ct= x.

7.2.1 =20 Address Context

7.2.2 =20 Document Context

7.2.3 =20 Email Context

7.2.4 =20 Numbers Context

7.2.5 =20 Unknown Context

7.2.6 =20 Normal Context

7.3 Mr= k.

7.4 Pa= u.

7.5 Pr= n.

7.6 Rm= S.

7.7 Rm= W...

7.8 RS= pd.

7.9 Rs= t

7.10 Spd.

7.11 Vce.

7.12 Vol

8 Microsof= t SAPI=20 5.1 Control Tags.

8.1 =20 ATT_Div.

8.2 =20 ATT_Ignore_case.

8.3 =20 Bookmark.

8.4 =20 Context

8.4.1 =20 Address Context

8.4.2 =20 Address_postal Context

8.4.3 =20 ATT_Literal Context

8.4.4 =20 ATT_Math Context

8.4.5 =20 ATT_Measurement Context

8.4.6 =20 Currency Context

8.4.7 =20 Date_MD Context

8.4.8 =20 Date_MDY Context

8.4.9 =20 Date_Year Context

8.4.10 =20 Number_Cardinal Context

8.4.11 =20 Number_Decimal Context

8.4.12 =20 Number_Fraction Context

8.4.13 =20 Phone_Number Context

8.4.14 Ti= me=20 Context

8.4.15 We= b=20 Context

8.4.16 We= b_url=20 Context

8.4.17 Em= ail=20 Context

8.4.18 =20 Email_address Context

8.5 Pr= on.

8.6 Ra= te.

8.7 =20 Silence.

8.8 Sp= ell

8.9 Vo= ice.

8.10 Volume.

9 SSML Con= trol=20 Tags.

9.1 =20 ATT_Ignore_Case.

9.2 Br= eak.

9.3 Ma= rk.

9.4 =20 Paragraph.

9.5 =20 Phoneme.

9.6 =20 Prosody.

9.6.1 =20 Rate.

9.6.2 =20 Volume.

9.7 =20 Say-As.

9.7.1 =20 Acronym..

9.7.2 =20 Address.

9.7.3 =20 ATT_Literal

9.7.4 =20 ATT_Math.

9.7.5 =20 ATT_Measurement

9.7.6 =20 Currency.

9.7.7 =20 Date.

9.7.8 =20 Name.

9.7.9 =20 Net

9.7.10 =20 Number

9.7.11 Su= b.

9.7.12 =20 Telephone.

9.7.13 Ti= me.

9.8 =20 Sentence.

9.9 Sp= eak.

9.10 Voice.

10 JSML=20 Control Tags.

10.1 jsml

10.2 div.

10.3 voice.

10.4 sayas.

10.4.1 =20 literal

10.4.2 da= te.

10.4.3 ti= me.

10.4.4 na= me.

10.4.5 ph= one.

10.4.6 ne= t

10.4.7 =20 address.

10.4.8 =20 currency.

10.4.9 =20 measure.

10.4.10 =20 number

10.5 phoneme<= SPAN=20 style=3D"DISPLAY: none; COLOR: windowtext; TEXT-DECORATION: none; text-und= erline: none; mso-hide: screen">.

10.6 break.

10.7 prosody<= SPAN=20 style=3D"DISPLAY: none; COLOR: windowtext; TEXT-DECORATION: none; text-und= erline: none; mso-hide: screen">.

10.7.1 ra= te.

10.7.2 =20 volume.

10.8 marker

10.9 engine.

11 =20 Custom Dictionaries.

11.1 Defining= Custom=20 Pronunciations.

11.2 Using th= e Win32=20 WinDictEdit Tool

11.2.1 De= fining=20 Replacements.

11.2.2 Ch= anging=20 the default pronunciation of a word.

11.3 Adding C= ustom=20 Pronunciations to Your Application.

12 =20 Performance Guidelines.

12.1 TTS Memo= ry=20 Requirements.

12.2 Definiti= ons.

12.3 Performa= nce=20 Results.

12.4 =20 Recommendations.

Appendix A: = Phonetic=20 Alphabets.

A.1. IPA Pho= netic=20 Alphabets.

A.1.1 US Eng= lish=20 SAPI 4/IPA Phonetic Alphabet

A.1.2. Spani= sh SAPI=20 4/IPA Phonetic Alphabet

A.1.3. Germa= n SAPI=20 4/IPA Phonetic Alphabet

A.1.4 French= SAPI=20 4/IPA Phonetic Alphabet

A.1.5 UK Eng= lish=20 SAPI 4/IPA Phonetic Alphabet

A.2.1 The SA= PI 5 and=20 DARPA US English Phonetic Alphabets.

A.2.2. SAPI = 5 SAMPA=20 Spanish Phonetic Alphabet

A.2.3. SAPI = 5 SAMPA=20 German Phonetic Alphabet

A.2.4. SAPI = 5 SAMPA=20 French Phonetic Alphabet

A.2.5. SAPI = 5 SAMPA=20 UK English Phonetic Alphabet

Appendix B: = SAPI 5.1=20 Compliance.

B.1. Support= for=20 SAPI 5.1.

1 = Introduction

1.1. = =20 AT&T Natural Voices Text-To-Speech

AT&T Natural Voices=20 Text-to-Speech (TTS) Engines provide synthesis services in multiple langua= ges=20 for application builders creating desktop applications. The application re= quests=20 services from the TTS engine using either a Microsoft SAPI 4 COM object, S= API=20 5.1 COM object, a Java program, or=20 using a convenient set of C++ APIs that are included with the product.=20 High-quality male and female voices are included in 8 KHz =B5law and CCITT G.711 alaw for telephony applicat= ions.=20 Additional voices and higher quality 16 KHz voices are available for=20 non-telephony applications. The AT&T Natural Voices TTS = product=20 is currently available in three configurations: the Server Edition, the Server-Lite Edition, and the Desktop Edition. The Server Edition pro= vides a=20 scalable, client/server architecture that=92s ideal for supporting many=20 simultaneous client connections to a single server or a large server farm = of TTS=20 servers. The Server-Lite Ed= ition=20 offers the same TTS technology, but on a single server where both the=20 application and the TTS run on a single computer making the Server-Lite Ed= ition=20 ideal for small office applications requiring fewer than about 10 simultan= eous=20 TTS channels[1]. The Desktop Edition provides the= same=20 AT&T Natural Voices TTS technology on a desktop computer but limits th= e TTS=20 engine to a single channel. = =20 All three configurations are intended for application developers to= use=20 to build TTS-enabled applications, either for scalable applications using = the=20 Server Edition, small to mid-sized applications using the Server-Lite Edit= ion,=20 or end user desktop applications using the Desktop Edition.

The following table shows = the=20 differences between the three editions of the AT&T Natural Voices=20 product.

Feature/Requirement<= /p>

Server

Server-Lite

Desktop

Architecture

Client/Server

Sin= gle=20 computer

Scalability

(simultaneous channels)

Unl= imited=20 channels across multiple TTS servers

Sev= eral=20 channels on one server

Sin= gle=20 channel

Design goals

Lar= ge channel=20 capacity

Sin= gle computer=20 with smaller memory requirements

Recommended&n= bsp;=20 Memory

512= MB

(25= 6=20 Minimum)

256= MB

(12= 8=20 Minimum)

256= MB

(12= 8=20 Minimum)

Minimum CPU

500= MHz=20 PIII

300= MHz=20 PIII

300= MHz=20 III

Disk Space

500= MB

Platforms

=A7 = =20 Windows NT, 2000, XP

=A7 = =20 Windows 98, ME

=A7 = =20 Linux 6.1, 6.2, 7.2

=A7 = =20 Solaris/SPARC 2.7, 2.8

=A7 = =20 HPUX 11.0

=A7 = =20 AIX 4.3

√

Cli= ent only

√

&n= bsp;

√

SNMP Support

√

Multi-Lingual

√

Voice Fonts

√

Development SDK and redistributable TTS engine<= /p>

√

SAPI 4.0

√

SAPI 5.0 & 5.1[2]

√

SSML

√

JSML

√

Custom dictionaries; Word and sentence notifica= tions;=20 user-defined bookmarks

√

This document describes the AT&T Natural Voices=20 Text-To-Speech Development SDK which application developers use to create= =20 TTS-enabled applications using US English, German, and Latin American Span= ish,=20 UK English, and French voices both for large capacity and desktop=20 environments. The SDK is th= e same=20 for all three editions, the only code difference being in creating the=20 engine.

[1]=20 The actual number of channels you may be able to support with the Server-L= ite=20 Edition depends upon the other applications that share the computer=92s pr= ocessing=20 power and memory. Text To S= peech=20 Synthesis tends to be memory intensive rather than CPU intensive so adding= =20 physical memory may allow you to run additional channels.

[2]=20 SAPI 5.0 does not support Spanish. = =20 You must upgrade to SAPI 5.1 to use the multi-lingual features of t= he=20 AT&T Natural Voices TTS 1.4 Release.