fundamentals speech mrcp

94
Fundamentals — Speech MRCP Avaya MS 7.0 NN44471-102, 02.01 3 Dec 2010

Upload: others

Post on 16-Jan-2022

24 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Fundamentals Speech MRCP

Fundamentals — Speech MRCP

Avaya MS 7.0NN44471-102, 02.01

3 Dec 2010

Page 2: Fundamentals Speech MRCP

© 2010 Avaya Inc.

All Rights Reserved.

Notice

While reasonable efforts have been made to ensure that theinformation in this document is complete and accurate at the time ofprinting, Avaya assumes no liability for any errors. Avaya reserves theright to make changes and corrections to the information in thisdocument without the obligation to notify any person or organization ofsuch changes.

Documentation disclaimer

“Documentation” means information published by Avaya in varyingmediums which may include product information, operating instructionsand performance specifications that Avaya generally makes availableto users of its products. Documentation does not include marketingmaterials. Avaya shall not be responsible for any modifications,additions, or deletions to the original published version ofdocumentation unless such modifications, additions, or deletions wereperformed by Avaya. End User agrees to indemnify and hold harmlessAvaya, Avaya's agents, servants and employees against all claims,lawsuits, demands and judgments arising out of, or in connection with,subsequent modifications, additions or deletions to this documentation,to the extent made by End User.

Link disclaimer

Avaya is not responsible for the contents or reliability of any linked Websites referenced within this site or documentation provided by Avaya.Avaya is not responsible for the accuracy of any information, statementor content provided on these sites and does not necessarily endorsethe products, services, or information described or offered within them.Avaya does not guarantee that these links will work all the time and hasno control over the availability of the linked pages.

Warranty

Avaya provides a limited warranty on its Hardware and Software(“Product(s)”). Refer to your sales agreement to establish the terms ofthe limited warranty. In addition, Avaya’s standard warranty language,as well as information regarding support for this Product while underwarranty is available to Avaya customers and other parties through theAvaya Support Web site: http://support.avaya.com. Please note that ifyou acquired the Product(s) from an authorized Avaya reseller outsideof the United States and Canada, the warranty is provided to you bysaid Avaya reseller and not by Avaya.

Licenses

THE SOFTWARE LICENSE TERMS AVAILABLE ON THE AVAYAWEBSITE, HTTP://SUPPORT.AVAYA.COM/LICENSEINFO/ AREAPPLICABLE TO ANYONE WHO DOWNLOADS, USES AND/ORINSTALLS AVAYA SOFTWARE, PURCHASED FROM AVAYA INC.,ANY AVAYA AFFILIATE, OR AN AUTHORIZED AVAYA RESELLER(AS APPLICABLE) UNDER A COMMERCIAL AGREEMENT WITHAVAYA OR AN AUTHORIZED AVAYA RESELLER. UNLESSOTHERWISE AGREED TO BY AVAYA IN WRITING, AVAYA DOESNOT EXTEND THIS LICENSE IF THE SOFTWARE WAS OBTAINEDFROM ANYONE OTHER THAN AVAYA, AN AVAYA AFFILIATE OR ANAVAYA AUTHORIZED RESELLER; AVAYA RESERVES THE RIGHTTO TAKE LEGAL ACTION AGAINST YOU AND ANYONE ELSEUSING OR SELLING THE SOFTWARE WITHOUT A LICENSE. BYINSTALLING, DOWNLOADING OR USING THE SOFTWARE, ORAUTHORIZING OTHERS TO DO SO, YOU, ON BEHALF OFYOURSELF AND THE ENTITY FOR WHOM YOU ARE INSTALLING,DOWNLOADING OR USING THE SOFTWARE (HEREINAFTERREFERRED TO INTERCHANGEABLY AS “YOU” AND “END USER”),AGREE TO THESE TERMS AND CONDITIONS AND CREATE ABINDING CONTRACT BETWEEN YOU AND AVAYA INC. OR THEAPPLICABLE AVAYA AFFILIATE (“AVAYA”).

Copyright

Except where expressly stated otherwise, no use should be made ofmaterials on this site, the Documentation, Software, or Hardwareprovided by Avaya. All content on this site, the documentation and theProduct provided by Avaya including the selection, arrangement anddesign of the content is owned either by Avaya or its licensors and isprotected by copyright and other intellectual property laws including thesui generis rights relating to the protection of databases. You may notmodify, copy, reproduce, republish, upload, post, transmit or distributein any way any content, in whole or in part, including any code andsoftware unless expressly authorized by Avaya. Unauthorizedreproduction, transmission, dissemination, storage, and or use withoutthe express written consent of Avaya can be a criminal, as well as acivil offense under the applicable law.

Third-party components

Certain software programs or portions thereof included in the Productmay contain software distributed under third party agreements (“ThirdParty Components”), which may contain terms that expand or limitrights to use certain portions of the Product (“Third Party Terms”).Information regarding distributed Linux OS source code (for thoseProducts that have distributed the Linux OS source code), andidentifying the copyright holders of the Third Party Components and theThird Party Terms that apply to them is available on the Avaya SupportWeb site: http://support.avaya.com/Copyright.

Trademarks

The trademarks, logos and service marks (“Marks”) displayed in thissite, the Documentation and Product(s) provided by Avaya are theregistered or unregistered Marks of Avaya, its affiliates, or other thirdparties. Users are not permitted to use such Marks without prior writtenconsent from Avaya or such third party which may own the Mark.Nothing contained in this site, the Documentation and Product(s)should be construed as granting, by implication, estoppel, or otherwise,any license or right in and to the Marks without the express writtenpermission of Avaya or the applicable third party.

Avaya is a registered trademark of Avaya Inc.

All non-Avaya trademarks are the property of their respective owners,and “Linux” is a registered trademark of Linus Torvalds.

Downloading Documentation

For the most current versions of Documentation, see the AvayaSupport Web site: http://support.avaya.com.

Contact Avaya Support

Avaya provides a telephone number for you to use to report problemsor to ask questions about your Product. The support telephone numberis 1-800-242-2121 in the United States. For additional supporttelephone numbers, see the Avaya Web site: http://support.avaya.com.

2 Fundamentals — Speech MRCP 3 Dec 2010

Page 3: Fundamentals Speech MRCP

Contents

Chapter 1: New in this release.................................................................................................7Features............................................................................................................................................................7

Support for Loquendo Speech Server......................................................................................................7Other changes...................................................................................................................................................7

Chapter 2: Introduction.............................................................................................................9

Chapter 3: MRCP fundamentals.............................................................................................11MRCP..............................................................................................................................................................11Speech recognition.........................................................................................................................................12Grammars.......................................................................................................................................................12Speech synthesis............................................................................................................................................13MRCP servers and clients...............................................................................................................................13Voice Extensible Markup Language................................................................................................................14Co-residency deployment...............................................................................................................................14Pooling............................................................................................................................................................15Protocols and standards.................................................................................................................................18Avaya supported MRCP parameters...............................................................................................................19Avaya implemented MRCP session licensing.................................................................................................19Avaya implemented MRCP DESCRIBE health check.....................................................................................19MRCP Franchising..........................................................................................................................................20

Chapter 4: MRCP operational fundamentals for Avaya MS.................................................21MRCP configuration through EM....................................................................................................................21Media speech resources.................................................................................................................................22Parameters and configuration.........................................................................................................................23MRCP supported languages...........................................................................................................................24Troubleshooting common problems................................................................................................................25

Chapter 5: IBM WebSphere Voice Server fundamentals......................................................27Product overview.............................................................................................................................................27Unsupported IBM WVS features.....................................................................................................................27Vender documentation reference material......................................................................................................28Security considerations...................................................................................................................................29Configuration considerations...........................................................................................................................30Port configuration............................................................................................................................................30Port number configuration...............................................................................................................................31Common ports on the Media Server node......................................................................................................32Extend the time an LVR or TTS session can remain idle................................................................................32Voice configuration..........................................................................................................................................33Disable unnecessary installed voices.............................................................................................................33Change default voice......................................................................................................................................34Administrative considerations.........................................................................................................................34Media Server management and configuration................................................................................................35Change the Media Server Node host name....................................................................................................36WebSphere Application Server host name and node name...........................................................................37Backup and restore.........................................................................................................................................38Vendor logging................................................................................................................................................39Media Server remote support..........................................................................................................................40

Fundamentals — Speech MRCP 3 Dec 2010 3

Page 4: Fundamentals Speech MRCP

Grammars and lexicons..................................................................................................................................40IBM WVS tools................................................................................................................................................41

Chapter 6: Nuance Speech Server fundamentals................................................................43Product overview.............................................................................................................................................43Avaya unsupported vendor functionality.........................................................................................................44Vendor documentation reference material......................................................................................................44Licensing considerations.................................................................................................................................45Configuration required on the Nuance Speech Server...................................................................................45Configure the NSSserver.cfg file.....................................................................................................................45Configure the NSSserver.cfg file for co-residency..........................................................................................47Configure the Baseline.xml file........................................................................................................................48Disable real time antivirus scanning on the Nuance Speech Server..............................................................48Remote support...............................................................................................................................................49User dictionary and the say as tag..................................................................................................................49User dictionary configuration...........................................................................................................................51Lexicons..........................................................................................................................................................51Say-as tags.....................................................................................................................................................52

Chapter 7: IBM WebSphere Voice Server installation and uninstallation..........................53IBM WebSphere Voice Server installation.......................................................................................................53Installation time...............................................................................................................................................57Installing IBM WVS 6.1 software.....................................................................................................................57Installing the Application Server Network Deployment 6.1.............................................................................57Installing the HTTP Server 6.1........................................................................................................................58Installing the WebSphere Voice Server...........................................................................................................59Verifying the software installation....................................................................................................................60Posting the installation procedure...................................................................................................................61Installing additional third-generation languages..............................................................................................61Installing fifth-generation TTS voices..............................................................................................................62Checking the fifth-generation TTS voice install...............................................................................................63IBM WebSphere Voice Server installation documentation..............................................................................64IBM WebSphere Voice Server uninstallation...................................................................................................64Uninstalling IBM WVS 6.1 software................................................................................................................65Uninstalling additional voices..........................................................................................................................65Uninstalling Web server plug-ins for WebSphere Application Server.............................................................67Uninstalling IBM HTTP Server........................................................................................................................67Uninstalling Application Server.......................................................................................................................68IBM WebSphere Voice Server host name change..........................................................................................68IBM WebSphere Voice Server host name change procedures.......................................................................69Changing the host name in the WebSphere Application server......................................................................71Changing the host name on IBM HTTP server...............................................................................................72Changing the host name in WebSphere Voice Server with ifix IC51934........................................................72Updating the WebSphere Application Server Web Server plug-in..................................................................73Updating the Administrative Console Startup Link with the new host name...................................................74

Chapter 8: Nuance Media Server installation and uninstallation........................................75Nuance Media Server installation...................................................................................................................75Installing the Nuance Media Server................................................................................................................75Nuance Media Server installation...................................................................................................................76Installing the Nuance Recognizer 9 software..................................................................................................78Installing the Nuance Recognizer Language model........................................................................................79

4 Fundamentals — Speech MRCP 3 Dec 2010

Page 5: Fundamentals Speech MRCP

Installing RealSpeak RS 4.5...........................................................................................................................79Prerequisites to installing RealSpeak RS 4.5..................................................................................................79Installing the RS 4.5 voices.............................................................................................................................80Installing the Nuance License manager..........................................................................................................81Installing Nuance Speech Server 5.x software...............................................................................................81Installing the license file..................................................................................................................................83Prerequisites to installing the license file........................................................................................................83Restarting the Nuance Speech Server............................................................................................................84Uninstalling the Nuance media server............................................................................................................84

Chapter 9: Nuance Vocalizer installation..............................................................................87Nuance Vocalizer installation..........................................................................................................................87Installing Nuance Vocalizer software..............................................................................................................87Installing the Nuance Vocalizer voices............................................................................................................88Installing the Nuance License Manager..........................................................................................................88Installing Nuance Speech Server software.....................................................................................................89Configuring Nuance Speech Server software.................................................................................................90Testing Nuance Speech Server installation.....................................................................................................92Example of the simple VXML application........................................................................................................92

Index.........................................................................................................................................93

Fundamentals — Speech MRCP 3 Dec 2010 5

Page 6: Fundamentals Speech MRCP

6 Fundamentals — Speech MRCP 3 Dec 2010

Page 7: Fundamentals Speech MRCP

Chapter 1: New in this release

The following sections detail what's new in Avaya Media Server Fundamentals — Speech MRCP,NN44471-102 for Avaya Media Server.

• Features on page 7

• Other changes on page 7

FeaturesAvaya Media Server release 7.0 contains the following new features:

Support for Loquendo Speech ServerAvaya Media Server release 7.0 provides support for Loquendo MRCP Speech Server.Loquendo Speech Server is based on MRCPv1 (Media Resource Control Protocol) and offersASR (Automatic Speech Recognition) and TTS (Text-To-Speech) capabilities.

Refer to Loquendo documentation for installing, licensing and configuring the LoquendoSpeech Server.

Node architectureLoquendo Speech Server must be installed on a dedicated server (node) and must not co-reside with Avaya Media Server. Avaya recommends that the Loquendo Speech Server nodebe located on the private LAN to avoid latency issues.

SecurityIt is important that a site-specific security plan is in place where the Avaya Media Server andLoquendo Speech Server nodes are located. For example, you need to know the ports thatmust remain open on the Avaya Media Server node.

Other changesThere are no other changes for this release.

Fundamentals — Speech MRCP 3 Dec 2010 7

Page 8: Fundamentals Speech MRCP

New in this release

8 Fundamentals — Speech MRCP 3 Dec 2010

Page 9: Fundamentals Speech MRCP

Chapter 2: Introduction

This document provides fundamental information about speech synthesis and speech recognition servicesfor the Media Applications Server 7.0 products.

Prerequisites

There are no prerequisites for this book.

Navigation

• MRCP fundamentals on page 11

• MRCP operational fundamentals for Avaya MS on page 21

• IBM WebSphere Voice Server fundamentals on page 27

• Nuance Speech Server fundamentals on page 43

• IBM WebSphere Voice Server installation and uninstallation on page 53

• Nuance Media Server installation and uninstallation on page 75

• Nuance Vocalizer installation on page 87

Media Resource Control Protocol (MRCP) provides speech services on the Avaya MS, with speechresources residing on the IBM WebSphere Voice Server 6.1 (WVS 6.1) or on the Nuance Speech Server(NSS).

This book is designed to explain the following:

• How MRCP fundamental concepts and other protocols that apply to speech services function withinthe Avaya MS environment

• How speech services on the IBM WVS 6.1 and the NSS Server function

• How to install and uninstall speech resources on the IBM WVS 6.1 and the NSS

Two versions of MRCP are in general use: MRCPv1 (MRCP) and MRCPv2. Avaya MS supports MRCPv1.

For additional information on the Avaya MS product see Avaya Media Server Commissioning,NN44471-301, Avaya Media Server Configuration – VoiceXML and CCXML Application Programming,NN44471-501, and Avaya Media Server Administration - Application Management, NN44471-601.

For information about fault management and alarms, see Avaya Media Server Fault Management,NN44471-700.

Fundamentals — Speech MRCP 3 Dec 2010 9

Page 10: Fundamentals Speech MRCP

Introduction

10 Fundamentals — Speech MRCP 3 Dec 2010

Page 11: Fundamentals Speech MRCP

Chapter 3: MRCP fundamentals

Media Resource Control Protocol (MRCP) allows a speech server to provide speech services like voicerecognition or voice synthesis to clients in the Media Applications Server 7.0 (Avaya MS 7.0) environment.Use the following information to answer questions you have about MRCP, MRCP servers and clients,protocols and standards, and Avaya-supported MRCP parameters.

Navigation

• MRCP on page 11

• MRCP servers and clients on page 13

• Protocols and standards on page 18

• Avaya supported MRCP parameters on page 19

• Voice Extensible Markup Language on page 14

• Co-residency deployment on page 14

• Pooling on page 15

• Avaya implemented MRCP session licensing on page 19

• Avaya implemented MRCP DESCRIBE health check on page 19

• MRCP Franchising on page 20

MRCPMedia Resource Control Protocol (MRCP) unifies resource messaging and controls in theMedia Applications Server environments. Use the following information when you work withthe Avaya implementation of the MRCP Media client and to answer questions you have aboutMRCP in the Avaya MS environment.

MRCP is a communication protocol that allows a distributed speech server to provide speechservices like recognition or synthesis to its clients. It's a text-based protocol that is sent overanother transport and MRCP is defined by the Internet Engineering Task Force (IETF). Twoversions are in use - MRCPv1 and MRCPv2 - with messages sent over Real Time StreamingProtocol (RTSP) in MRCPv1 and Session Initiation Protocol (SIP) in MRCPv2. In both versions,the media is transported over Secure Real-time Transport Protocol (S)RTP.

Fundamentals — Speech MRCP 3 Dec 2010 11

Page 12: Fundamentals Speech MRCP

Avaya MS currently supports the following MRCPv1 servers:

• Nuance Speech Server 5.0.x

• Nuance Recognizer 9.0.y

• Nuance RealSpeak 4.5 patch 1

• IBM WVS 6.1.1

Speech recognitionSpeech recognition is the technology that matches spoken input to the expected input (orgrammar) in a speech application. Use the following information to answer general questionsabout speech recognition in the Avaya MS 7.0 environment.

Speech recognition (Automatic Speech Recognition or ASR and Large VocabularyRecognition or LVR) is the process of converting human speech to text. Most often ASRengines use specific grammars to narrow the possible matches for an audio segment. Mostengines include an end pointer that determines the beginning and end of speech, and thesesignal analyzers train themselves to the acoustic characteristics of the audio channel torecognize speech and filter noise. After the end pointer has isolated the audio segment, it sendsthe segment to the recognition engine for analysis.

Avaya MS uses MRCP servers (IBM and Nuance) to support speech recognition.

GrammarsGrammars are the expected spoken input in a speech application and are classified as staticor dynamic.

In the Avaya MS environment, static grammars are those grammars that load when the MRCPresource is allocated by the Avaya MS.

Static grammars cannot be loaded during startup because the TCP control connection betweena MRCP Client and a Media Server is only established during a SETUP operation. The SETUPoperation occurs when a request is made for a resource at the time a call is received.

Dynamic grammars are those grammars that the application sends to the recognizer during acall, using the DYNAMIC-GRAMMAR method. They are associated with a Uniform ResourceIdentifier (URI), a URI with ECMA, or a grammar string that the application builds based uponinput about the caller.

Avaya recommends that large grammars or grammar strings be precompiled and loaded onthe Media Server, where the preloading feature is available. Callers experience a delayotherwise.

MRCP fundamentals

12 Fundamentals — Speech MRCP 3 Dec 2010

Page 13: Fundamentals Speech MRCP

Some vendors support preloaded grammars by using a recognition server configuration file.However, this is outside of the MRCP specification and the MRCP Client does not know whichgrammars are loaded on the Media Server. A DYNAMIC-GRAMMAR message is sent to themedia servers to load grammars.

Grammars are unloaded when the application releases the LVR resource based on the cachemechanism that the Media Server uses. Unloaded dynamic grammars are not static.

Speech synthesisSpeech synthesis or Text-To-Speech (TTS) is the technology that allows a server to talk to acustomer in the Media Applications Server 7.0 (Avaya MS 7.0) environment. The followinginformation focuses on TTS resources on the Media Applications Server 7.0 (Avaya MS 7.0)products.

TTS is the process of converting text to computer-generated speech. Implementation of TTSalgorithms varies, but many implementations use pure audio synthesis to generate theresulting waveform. Others use waveform concatenation that pieces together tiny segmentsof prerecorded speech to build a complete waveform. These engines are often more resourceintensive when compared to pure synthesis engines.

Avaya MS uses the MRCP servers (IBM and Nuance) to stream TTS, as the TTS synthesizes inreal time.

MRCP servers and clientsThird party servers provide speech recognition and speech synthesis services. The followinginformation discusses Media Resource Control Protocol (MRCP) servers and clients on theMedia Applications Server (Avaya MS).

Third party MRCP servers provide speech capabilities and the MRCP servers are accessedvia MRCP, using an embedded MRCP client stack. A single MRCP server provides synthesisand recognition resources, if this configuration is supported by the vendor.

Avaya MS communicates with the MRCP server using an embedded MRCP client in the IvrMP.The MRCP server deploys as a co-resident or as a pooled network resource. A MRCP serverprovides synthesis and recognition resources if the vendor supports this configuration.

For more information about vendor-supported synthesis and recognition capabilities, see Table1: Synthesis and recognition capabilities on page 14.

Speech synthesis

Fundamentals — Speech MRCP 3 Dec 2010 13

Page 14: Fundamentals Speech MRCP

Table 1: Synthesis and recognition capabilities

Vendor Co-residency withAvaya MS

LVR TTS

Nuance Recognizer9.0

Yes Yes No

RealSpeak 4.5 Yes No Yes

IBM WVS 6.1 No Yes Yes

Nuance Recognizer9.0 and RealSpeak4.5 co-res

Yes Yes Yes

Voice Extensible Markup LanguageDevelopers use Voice Extensible Markup Language (VXML) to create Interactive VoiceResponse (IVR) applications on the Media Applications Server 7.0 (Avaya MS 7.0). Thefollowing information focuses on VXML and administrators can read this information beforeworking with IVR resources.

Avaya MS 7.0 executes applications written in VXML and fetches these application from thelocal file system or a remote Web server using HTTP. VXML applications are capable ofcreating interactive audio dialogs using prerecorded audio, synthesized speech, and inputusing Dual-tone Multifrequency (DTMF) or speech, or DTMF and speech.

The Avaya MS VXML browser is compliant with the W3C VoiceXML 2.0 Recommendation 16March and VoiceXML 2.1 2004 Recommendation 19 June 2007.

VoiceXML applications support UTF-8 and ISO-8859-1 based languages for speechrecognition and speech synthesis.

You can configure the debug parameters for VXML using the Debug For Voice XML Interpreterfeature of the Element Manager (EM).

For additional information, see Avaya Media Server Configuration – VoiceXML and CCXMLApplication Programming, NN44471-501.

Co-residency deploymentThe Media Applications Server 7.0 (Avaya MS 7.0) supports co-residency (co-res) deploymentof Media Resource Control Protocol (MRCP) . The following information focuses on co-resdeployment and it should be read before you work with resources in a co-res deployment.

MRCP fundamentals

14 Fundamentals — Speech MRCP 3 Dec 2010

Page 15: Fundamentals Speech MRCP

Administrators can install an MRCP server so that the server is co-res on the same COTShardware as Avaya MS. In this configuration the MRCP server resources are dedicated toAvaya MS.

This configuration impacts capacity and is only supported if the MRCP vendor supports a co-res configuration.

Co-res deployment is not supported by IBM WVS. But NSS supports co-res configuration.

See the appropriate vendor documentation for the use, features, and compatibility of allCommercial-Off-The-Shelf (COTS) hardware.

PoolingThe Media Applications Server 7.0 (Avaya MS 7.0) supports pooling of Media ResourceControl Protocol (MRCP) servers across one or more Avaya MSs to maximize speechresources. The following information focuses on pooling in the Avaya MS 7.0 environment.Use this information when working with pooled servers.

Each Avaya MS server can define one or more MRCP server pools that are used to allocatespeech resources. Each pool is named and contains a list of servers of the same type. Allservers in the same pool must be from the same vendor and resource type (for example, TTS orLVR). In addition all servers must share at least one language in common. Languagessupported for the pool are the languages that are common across all servers in that pool. ThePool Name and Pool Type values cannot be modified after the pool has been created.

Pooling is not allowed in a co-resident standalone environment.

During resource allocation, the pool name, required attributes (for example, language or type),or both are specified explicitly by the application or implicitly by the VXML interpreter on behalfof the application. Avaya MS uses the pool name and attributes to determine what pool canbe used from which to select a MRCP server. The end user marks a pool as deactivated,meaning it's not available for allocation to the application. To make it available you must mark itas activated.

After a pool is selected, the Avaya MS establishes a MRCP session to a server in the pool,using a weighted Round Robin algorithm of all resources in the pool that match the specifiedattributes.

The end user marks the MRCP servers as locked or the Avaya MS 7.0 detects them as down. Alocked or down server is temporary removed from the pool selection candidate list.

The algorithm selects the least weighted resource in the least weighted matching pool, unlessyou select a default pool. The algorithm determines the least weighted resource by multiplyingthe resource weight by the number of active sessions from the current server, selecting theresource with the least total weight.

Pooling

Fundamentals — Speech MRCP 3 Dec 2010 15

Page 16: Fundamentals Speech MRCP

Proper configuration of a resource weight allows you to group resources with differentcapacities into the same pool, without the applications overloading one or more servers. A poolweight is computed by multiplying the summed weight of its resource by the weightedconfiguration value of a pool. When you do not indicate a default pool, the algorithm selectsthe pool with the lowest summed weight.

If you configure multiple Avaya MS nodes to use the same MRCP resource, the session fromone node does not impact the weighting computation of the other node. But each node isunaware that the other is using the MRCP server and this can lead to a pool becomingoverloaded.

Every five minutes, Avaya MS 7.0 does a status check on unused resources. This checkincludes the allocation and deallocation of an inactive resource to determine if the resource ishealthy and available. If a resource fails a status check, it is deactivated and the platform issuesthe appropriate alarms and logs.

A deactivated resource reactivates when it passes a status check.

Element Manager (EM) provides a MRCP configuration section that allows you to managepooled servers. See MRCP configuration through EM on page 21 for more information.

MRCP fundamentals

16 Fundamentals — Speech MRCP 3 Dec 2010

Page 17: Fundamentals Speech MRCP

Figure 1: Pooled MRCP server

Pooling

Fundamentals — Speech MRCP 3 Dec 2010 17

Page 18: Fundamentals Speech MRCP

Protocols and standardsServers communicate with each other using protocols and standards. Use this informationwhen you have questions about these standards in the Media Applications Server 7.0 (AvayaMS 7.0) environment.

Avaya MS 7.0 uses the following protocols to facilitate speech recognition and speechsynthesis:

• Media Resource Control Protocol (MRCP)

• Real-time Streaming Protocol (RTSP)

• Real-time Transport Protocol (RTP)

• Speech Synthesis Markup Language (SSML)

• Semantic Interpretation for Speech Recognition (SISR)

• Speech Recognition Grammar Specification (SRGS)

In the Avaya MRCP implementation, Avaya MS 7.0 creates and breaks down a session foreach call, using the following protocols:

• RTSP (Real-time Streaming Protocol) for initiating and tearing down sessions with aMedia Server

• RTP (Real-time Transport Protocol) to send audio data to, and receive audio data from,the Media Server

Each speech vendor supports a different standard. For more information, refer to the vendordocumentation.

If an application has allocated an MRCP synthesis resource and an MRCP recognitionresource, the application has allocated two separate resources. However, they interact withthe same Media Server for processing purposes. This means that the SETUP (allocate or get)and TEARDOWN (deallocate or free) for each of these resources is separate and occursindependently of each other.

The MRCP libraries include three software modules (or stacks) that implement the MRCP,RTSP, and RTP protocols. Each library contains the following software modules:

• MRCP software module that implements the MRCP protocol

• RTSP software module that implements the RTSP protocol

• RTP software module that implements the RTP protocol

In addition, MRCP relies on Speech Synthesis Markup Language (SSML) to convert writtentext to computer generated speak. For more information, see Speech Synthesis MarkupLanguage (SSML) 1.0.

MRCP relies on Semantic Interpretation for Speech Recognition (SISR) - acting in conjunctionwith Speech Recognition Grammar Specification (SRGS) - to decorate the grammars with tag

MRCP fundamentals

18 Fundamentals — Speech MRCP 3 Dec 2010

Page 19: Fundamentals Speech MRCP

elements. For more information, see Semantic Interpretation for Speech Recognition (SISR)- W3C Working Draft 1 April 2003.

Avaya supported MRCP parametersMedia Resource Control Protocol (MRCP) parameters are values that you can set to maximizespeech resources on the Media Applications Server 7.0 (Avaya MS 7.0). The followinginformation focuses on Avaya-supported parameters. Use the following information whenworking with these values.

Avaya MS 7.0 supports the MRCPv1 specification.

Avaya MS 7.0 does not support:

• the recognition Large-Vocabulary-Recognition (LVR) MRCPv1 method GET_RESULT.

• the synthesis (TTS) MRCP method PAUSE, RESUME, and CONTROL.

Avaya MS 7.0 does support SPEECH-MARKER events. But some vendors do not supportSpeech marker events with MRCPv1. Based on Avaya testing of vendor resources:

RealSpeak supports SPEECH-MARKER with numeric mark names. For example, 12345,111, or 344.

Avaya implemented MRCP session licensing

Important:MRCP session licensing only applies to Avaya's implementation of MRCP clients and doesnot pertain to the licensing of the media server.

Each MRCP session in use by the system requires one mrcp::sess license from the licenseserver. For example, a session using both TTS and LVR requires two licenses. The licensesare taken when a resource is allocated and returned when a resource is released. Failure ofan MRCP resource causes all related licenses to be returned to the license server. Themrcp::sess license, from the perspective of the license server and local license pooling,behaves the same as all other license types.

Avaya implemented MRCP DESCRIBE health checkThe DESCRIBE request for RTSP (Real-Time Streaming Protocol RFC 2326) is implementedas a mechanism for sampling the health of configured MRCP resources. The DESCRIBE

Avaya supported MRCP parameters

Fundamentals — Speech MRCP 3 Dec 2010 19

Page 20: Fundamentals Speech MRCP

request replaces use of the SETUP request which required a speech server license to function.The new mechanism sends a DESCRIBE request to idle the MRCP media server according tothe configured Resource Ping Interval. A functioning MRCP media server generates repliesthat indicate that the resource is healthy and providing service. An alarm is raised if a resourcefails to respond and a corresponding event log indicates each failed resource. MRCP resourcesthat are actively in-use are not checked.

The MRCP resource health check is disabled by setting the MRCP Resource Ping Intervalto 0. Disabling the heath check is not recommended. When disabled, failed resources are notremoved from the set of active servers. Instead, they remain part of the selection set whichcan introduce delays as failed resources are tried, and re-tried, until a functioning resource isidentified. When the health check is enabled, failed resources are not selected until theyrespond to a health check. The system tries to recover failed resources about once a second inorder to recover them as soon as they return to service. A resource that is planned to be outof service should be locked or removed from its pool(s) in order to avoid raising an alarm.

Allocation of an MRCP resource begins with the acquisition of an mrcp::sess license. Therequest fails if a license cannot be acquired. Once a license is acquired, the system selectsthe least loaded active resource that matches the request criteria. If allocation from a resourcefails (assuming that the resource ping is enabled) the system marks the related resource asfailed and attempts to allocate from up to the next two least loaded resources (according tothe setting of the MRCP Retry Limit). The allocation sequence returns a failure to the serviceif an allocation does not succeed before the retry limit is reached.

MRCP FranchisingA task in the MRCP section in Element Manager known as Custom MRCP Vendors allowsadministrators to add custom vendors with assigned resource types (caps) and the list oflanguages available to each cap. Note, administrators may add, edit and delete Custom MRCPVendors on a primary server only.

For additional information, see Avaya Media Server Commissioning, NN44471-301.

MRCP fundamentals

20 Fundamentals — Speech MRCP 3 Dec 2010

Page 21: Fundamentals Speech MRCP

Chapter 4: MRCP operational fundamentalsfor Avaya MS

The following information provides help to common issues, including Media Resource Control Protocol(MRCP) configuration through Element Manager (EM), performance, fault monitoring and troubleshootingcommon problems.

Navigation

• MRCP configuration through EM on page 21

• Troubleshooting common problems on page 25

MRCP configuration through EMElement Manager (EM) allows administrators to configure the Media Resource ControlProtocol (MRCP) servers to maximize speech resources. The following information focuseson using EM to configure MRCP servers. Read this information before you work with theseresources.

EM has a MRCP configuration section with the following capabilities:

• Adding and deleting pools including adding and removing MRCP servers to and frompools, and assigning weights to servers in a pool. When adding servers to pools EMensures that all servers in the pool are the same type, from same vendor, and have atleast one language in common.

• Specifying one default TTS and LVR pool.

• Displaying a list of current pools defined, display the languages supported by each pool,and the MRCP servers associated with each pool.

• Adding and deleting MRCP servers. Also the ability to edit each MRCP server propertiesand attributes. For example, vendor, type, or supported languages.

For more information about the available MRCP settings, see Table 2: MRCP generalsettings on page 21.

Table 2: MRCP general settings

MRCP settings DefinitionMaximum MRCP Channels The maximum number of MRCP channels

allocated by the Media Applications Server

Fundamentals — Speech MRCP 3 Dec 2010 21

Page 22: Fundamentals Speech MRCP

MRCP settings Definition7.0 (Avaya MS 7.0). The default value is 100.The range is 0 to 2 000.

Retry Limit The number of retry attempts when resourceallocation fails and the value range is 0 to 2.The least loaded server is tried first, followedby the second least loaded, and the thirdleast loaded. The default value is 2.

MRCP Resource Ping Interval The number of seconds between successivetest allocations of configured MRCPresources. A value of 0 disables testallocations. Default value is 300. The rangeis 0 to 31 536 000.

MRCP RTP Start Port The local RTP start port to use for MRCPv1RTP connections to speech servers. Defaultvalue is 5 554. The range is 1 024 to 64 000.

Transaction Timeout The maximum duration (ms) that an MRCPtransaction can last before being terminatedand a resource fault generated. The range is0 to 31 536 000. The default value is 10 000.

Recognition Timeout The maximum duration (ms) that an MRCPtransaction can last before being terminatedand a resource fault generated. The range is0 to 31 536 000. The default value is 600 000.

For additional information see Avaya Media Server 6.1 Commissioning

Media speech resourcesThird party MRCP servers provide speech synthesis and recognition capabilities. However,the Avaya MS environment provides users with the capability to configure the integrationbetween the MRCP server and the MRCP client. The following information focuses onconfiguring the speech resources.

MRCP provides the capability to configure the integration between the MRCP Server and theMRCP Client, and to group servers by pools. These servers provide speech capability to thenetwork of Avaya Media Server nodes. You can allocate speech resources by the application(or VXML browser), and by specifying a language, vendor or pool name.

All MRCP configuration and maintenance tasks are on the MRCP branch in Element Manager(EM) menu pane. The administrator can access the summary page by clicking on the MRCPtask in the menu pane. The summary page provides links to access the General Settings page,Servers page, and Pools page.

MRCP operational fundamentals for Avaya MS

22 Fundamentals — Speech MRCP 3 Dec 2010

Page 23: Fundamentals Speech MRCP

Administrators use the server resource page to configure the MRCP resource information.

For additional information see Avaya Media Server Commissioning, NN44471-301

Parameters and configurationUsing Element Manager (EM), administrators can configure speech parameters on the MediaResource Control Protocol (MRCP) servers. You can configure the following speechparameters following through EM. Use the following list of common parameters and theirdefinitions when working with speech resources.

MRCP servers provide speech resources. One or more speech servers must have networkaddress and capabilities configured for the Avaya MS to provide speech synthesis orrecognition capabilities to the application.

In EM, you can add or delete entire MRCP pools as well as add or remove individual MRCPservers from a pool. To see a list of servers defined in a pool, you log on to EM and follow thepath: System Configuration, Signaling Protocols, MRCP, Pools

Select the pool name and the right pane shows a list of servers added to the pool.

For additional information on using EM to configure parameters see Avaya Media ServerCommissioning, NN44471-301

For more information on MRCP server parameters for speech resources, see Table 3: MRCPserver parameters for speech resources on page 23.

Table 3: MRCP server parameters for speech resources

Parameter DescriptionServer Name Display name used to identify this MRCP server and is

present when defining and configuring MRCP pools.

Server Description Description of the server.

Server Address Host name or IP address of the MRCP server. Using anIP Address is recommended. If you enter a host name,you must activate the DNS in the network.

Port Port that MRCP server is listens for MRCP requests. Thedefault is 4554.

Vendor ID MRCP vendor identifier. The options are Nuance MediaServer or WebSphere Voice Server.

Codec Audio codec that can be used. The options are G.711ULaw, G.711 ALaw or PCMU. The default value is PCMU.

State Initial operational state of the MRCP server and can beeither active or deactive.

Parameters and configuration

Fundamentals — Speech MRCP 3 Dec 2010 23

Page 24: Fundamentals Speech MRCP

Parameter DescriptionType The MRCP resource capabilities supported, which can

be one of: LVR Only, TTS Only, or LVR and TTS.

Weight The pool weight is a value that is used to assign a loadlevel to a pool. A pool's load is determined by the numberof active resources it has times the weight. When a pool isselected from amongst a set of available pools, the poolhaving the lead load is chosen.

Max. Sessions Max Sessions is the total number of MRCP sessions thatcan be active in the node at one time. A call using bothTTS and LVR will consume 2 MRCP sessions.

URL Suffix The URL suffix that identifies the LVR resource. Thedefault is media/recognizer or media/synthesizer.

Languages The languages supported by this resource.

Table 4: MRCP pool parameters

Parameter DescriptionPool Name Name of the pool.

Pool description Description of the pool.

Pool type The type of pool, either TTS or LVR.

Vendor name The vendor name, either Nuance or IBM.

Assign servers A list of servers available to be assigned tothis pool.

Pool weight The relative amount of processing resourcesused by pools for the MRCP servers. Uselarger values for pools that have MRCPresources that use more processingresources per session.

Language Add languages supported by this resource.

Default pool The default pool is the pool selected when nospecific pool is identified by a request toreserve an MRCP resource.

MRCP supported languagesThe following information lists the common languages supported by Media Resource ControlProtocol (MRCP). Use it to answer any questions about MRCP supported languages.

MRCP operational fundamentals for Avaya MS

24 Fundamentals — Speech MRCP 3 Dec 2010

Page 25: Fundamentals Speech MRCP

MRCP supported languages are vendor specific. The languages supported depend on thecapabilities of the speech server. Languages can be added or removed on Element Manager.

For more information about MRCP supported languages, see Table 5: MRCP supportedlanguages on page 25.

Table 5: MRCP supported languages

Language Country VendorEnglish United States Nuance, IBM WVS

English Australian IBM WVS

English Great Britan IBM WVS

French France Nuance, IBM WVS

Español (Spanish) Spain Nuance, IBM WVS

Deutsch (German) Germany Nuance, IBM WVS

Japanese Japan Nuance, IBM WVS

Traditional Chinese Taiwan Nuance

Simplified Chinese China Nuance

Korean Korea Nuance

Portuguese Portugal Nuance

Russian Russia Nuance, IBM WVS

Troubleshooting common problemsAvaya Media Server allows you to troubleshoot common problems or issues. The followinginformation lists common alarms on the Avaya Media Server 7.0 (Avaya MS 7.0).

The corrective actions for each alarm are listed with the alarm descriptions.

See Avaya Media Server Troubleshooting, NN44471-702 for instructions to troubleshootcommon problems.

For more information about common alarms and logs for troubleshooting, see Table 6:Common alarms and logs for troubleshooting on page 26.

Troubleshooting common problems

Fundamentals — Speech MRCP 3 Dec 2010 25

Page 26: Fundamentals Speech MRCP

Table 6: Common alarms and logs for troubleshooting

Alarmsand logs

Message Action

Alarm 331 MRCP Resources Locked; All MRCPresources are locked.

Unlock or unprovision all lockedMRCP resources or associatedresource types.

Alarm 332 MRCP Resource locked; One or moreMRCP resources are locked.

Unlock or unprovision all lockedMRCP resources or unprovision allassociated resource types (aresource must have a defined typebefore it raises an alarm).

Alarm 334 MRCP Resource Failed (IvrMP); Oneor more MRCP resources have failedhealth checks.

This alarm clears when all unlockedMRCP resources respond to regularhealth checks. The failing resourcesare identified in the application log.Restore proper functionality of theseresources, or, deactivate or removethem from their pool.

Log 7044 MRCP Resource Failed This log has several forms. The logoccurs when an error is generated bya MRCP resource. The log identifiesthe failed resource, its session, andthe state of the resource when theerror occurred.

Log 7105 MRCP Allocation Failure Matching resources are in a failedstate [pool <pool name>, capability<LVR or TTS>, vendor <vendortype>, version <version data>], LOG<error information>. The <errorinformation> contains informationabout why the resource cannot beallocated. The error informationrecords what occurred when aresource selection failed. You usethis information, with the relatedMRCP resource provisioning in EM,to determine if the provisioning isincorrect or if additional resourcesmust be provisioned.

MRCP operational fundamentals for Avaya MS

26 Fundamentals — Speech MRCP 3 Dec 2010

Page 27: Fundamentals Speech MRCP

Chapter 5: IBM WebSphere Voice Serverfundamentals

IBM WebSphere Voice Server V6.1 (WVS 6.1) is a software product that you use with other software andhardware telephony products to provide speech resources such as speech recognition and speechsynthesis. The following information provides an overview of the IBM WVS 6.1 in the Media ApplicationsServer 7.0 (Avaya MS 7.0) environment.

Navigation

• Product overview on page 27

• Vender documentation reference material on page 28

• Security considerations on page 29

• Configuration considerations on page 30

• Administrative considerations on page 34

• IBM WVS tools on page 41

Product overviewIBM WebSphere Voice Server is a Media Resource Control Protocol (MRCP)-based speechresource that provides speech recognition and speech synthesis, including Automatic SpeechRecognition (ASR) and Text-To-Speech (TTS) capabilities. The following information is anoverview of the IBM WVS 6.1 and the role it plays in the Media Applications Server 6.4 (AvayaMS 6.4) environment.

Avaya supports the WVS 6.1 resource only on Microsoft Windows 2008 nodes.

IBM WVS 6.1 is supported on Microsoft Windows 2008 nodes with Service Pack 2 required.IBM only supports MRCPv1 and not MRCPv2.

Unsupported IBM WVS featuresIBM WebSphere Voice Server 6.1 (WVS 6.1) provides speech resources in the MediaApplications Server 7.0 (Avaya MS 7.0). The following provides information about features that

Fundamentals — Speech MRCP 3 Dec 2010 27

Page 28: Fundamentals Speech MRCP

Avaya does not support on the IBM WVS 6.1. Use this information to answer questions thatyou have about unsupported features, including features that are constrained.

The current Avaya implementation of IBM WVS 6.1 does not support the following features ofIBM WVS 6.1:

• WebSphere Application Server Edge Load Balancer

• load balancing from the IBM WVS is not supported

The messaging between the Media Control Protocol (MRCP) Client and MRCP Server is thesame regardless of the vendor. For this reason, all methods are implemented on the MRCPClient side. If a Media Server does not support a particular MRCP method, it fails that request.The Media Server fails an unknown request if the application sends an unknown or invalidrequest to the Media Server.

In addition, the following general support issues are known:

• The Avaya MRCP Client does not support the GetResult request.

• Avaya does not support the parameter.

• The Avaya MRCP Client does not support the Pause, Resume, and Control the speechoutput from the Media Server during a synthesis.

Vender documentation reference materialThe following information is a list of available IBM documentation for the IBM WebSphere 6.1(WVS 6.1). Use it for additional reference material on the IBM WVS.

The IBM WVS Information Center (InfoCenter) provides comprehensive information about allaspects of working with IBM WVS. Use the InfoCenter as a central reference resource. Formore information, go to the InfoCenter Web site: http://www.ibm.com/us/en/.

In addition to the InfoCenter, IBM has developed reference material that applies to tuning theIBM WVS. This material includes the following documents:

• For more information about IBM Redbooks, go the Redbooks Web site: http://www.redbooks.ibm.com

- IBM WebSphere Voice Server for Multiplatforms Handbook, SG24-6447-00. Themost comprehensive reference resides in Chapter 4 "Administration andConfiguration".

- Speech User Interface Guide, REDP-4106-00, contains references about RAD,Voice Toolkit, Building Grammars, Voice Trace Analyzer, and tuning VoiceXMLapplications.

• The WebSphere Voice Server Information Center (InfoCenter) provides information aboutthe following topics:

IBM WebSphere Voice Server fundamentals

28 Fundamentals — Speech MRCP 3 Dec 2010

Page 29: Fundamentals Speech MRCP

- WebSphere Voice Server http://www.ibm.com/us/en/.

Using the bookmarks, browse to Product Overview, Product Descriptionfor information about languages supported for Automatic Speech Recognition (ASR)and Text-To-Speech (TTS) and voices, genders, and lexicons supported for TTS.

- Tuning Windows Server 2008. http://www.ibm.com/us/en/

- Tuning WebSphere Voice Server. http://www.ibm.com/us/en/

- IBM WVS configuration parameters. http://www.ibm.com/us/en/

• For information about grammar cache control guidelines, see ASR Troubleshooting Tipsin the IBM WVS InfoCenter at the following link: http://www.ibm.com/us/en/

Other reference material include

• WebSphere Voice Server: Product Overview http://www.ibm.com/us/en/.

• IBM developer Works: WebSphere Voice Zone http://www.ibm.com/us/en/.

Security considerationsAvaya recommends that you keep IBM WebSphere Voice Server 6.1 (WVS 6.1) server in asecure environment. The following information outlines security issues that administratorsmust consider when installing and maintaining the IBM WVS 6.1. Use this information toanswer questions that you have on site-specific plans and security hardening on the IBM WVS6.1.

You must have a site-specific security plan in place and that this plan is available on site forthe IBM WVS 6.1. This plan must address security on the Media Server node and on theSpeech Server node. For example, administrators should be aware of the ports that mustremain open on the Media Server node and include this information in the security plan.

When considering security on the IBM WVS 6.1, administrators must be aware that the AvayaMedia Server 7.0 (Avaya MS 7.0) hardening disallows port 554. The Real-time TransportProtocol (RTP) stream locks this port.

To change the port number on IBM WVS 6.1 enter the following command. You must enter thiscommand from the D:\Program Files\WebSphere\VoiceServer\samples directory. putParamcom.ibm.voice.server.rtspbridge.port <port number>.

The path to appserver\bin directory for wvs must be set. The HTTP server, Administrationserver, and Application server for IBM WVS must be running when executing this command.Restart the server after changing the parameter.

Security considerations

Fundamentals — Speech MRCP 3 Dec 2010 29

Page 30: Fundamentals Speech MRCP

Configuration considerationsUse the following information for voice synthesis and voice recognition configuration on theIBM WebSphere Voice Server 6.1 (WVS 6.1).

The configuration relates to the IBM VWS Media Server, where the IBM WVS 6.1 vendorsoftware is installed. For more information, see Media Server management andconfiguration on page 35.

For more information about configuring the Media Server node, see the WebSphere VoiceServer Information Center (InfoCenter). The InfoCenter provides comprehensive informationabout all aspects of working with IBM WVS. Use this as a central reference resource.

Use the IBM Capacity Planning tool to determine the number of channels that an IBM MediaServer node can support for Large Vocabulary Recognition (LVR) or Text-To-Speech (TTS)engines.

You can start the ICM node with more channels than are configured on the IBM WVS 6.1 MediaServer node. A channel on the Avaya Speech Server is not associated with a speech synthesis(Text-To-Speech or TTS) or recognition (Large Vocabulary Recognition (LVR) and AutomaticSpeech Recognition (ASR) engine on the IBM WVS 6.1 Media Server node.

The IBM WVS Media Server manages the pool of LVR/ASR or TTS engines internally in such away that it can handle more simultaneous sessions than the number of LVR/ASR or TTSengines configured for use.

Port configurationUse the following information as an overview for configuring ports on the IBM WebSphereVoice Server 6.1 (WVS 6.1), including Automatic Speech Recognition (ASR) and Text-To-Speech (TTS) ports configuration with wsadmin scripting.

Before you can configure the WebSphere Application Server using the scripting interfacewsadmin, you must

• browse to the Websphere, VoiceServer, samples directory

• start the WebSphere Application Server by selecting Start, Programs, IBMWebSphere, Application Server V6.1, Start the Server

In the following scenario, assume you purchased 10 ports of Automatic Speech Recognition(ASR) and Text-To-Speech (TTS). Now, you want to change the ASR and TTS engine poolsizes to 10 for LVR and 10 for TTS.

IBM WebSphere Voice Server fundamentals

30 Fundamentals — Speech MRCP 3 Dec 2010

Page 31: Fundamentals Speech MRCP

Issue the following commands, one at a time:

• cd C:\Program Files\WebSphere\VoiceServer\samples• putParam.bat com.ibm.voice.server.recognizer.asr.pool-size 10true

• putParam.bat com.ibm.voice.server.synth.pool-size 10 true• putParam.bat com.ibm.voice.server.rr.fetchexpires 691200000 true

You must restart the Media Server node before these changes take effect.

To check that the new parameter values are in effect, change the command from put to get,and replace the parameter and value with the wildcard $ as in the following LVR and TTSexample:

getParam.bat com.ibm.voice.server.recognizer$You can perform management and monitoring operations on the Media Server node using theIBM WVS Administrative Console (AdminConsole) and the scripting interface %JAVA_HOME%\..\bin\wsadmin.bat.

• The AdminConsole is a Graphical User Interface (GUI) tool that you use to manageoperations on the Media Server node. The URL for the Administrative Console is:

http://<hostname>/9090:/admin/• The management operations you cannot accomplish with the AdminConsole, you can

accomplish using the scripting interface:

%JAVA_HOME%\..\bin\wsadmin.batFor more information about configuring the Media Server node, see Media Servermanagement and configuration on page 35.

Port number configurationUse the following information as an overview for configuring the port number on the IBMWebSphere Voice Server6.1 (WVS 6.1).

To change the port number on the 6.1 (WVS 6.1) use the following command:

putParam com.ibm.voice.server.rtspbridge.port <port number>.

The path to appserver\bin directory for the WVS 6.1 must be set. The HTTP server,Administration Server and Application Server for WVS must be running when executing thiscommand. Restart the MRCP speech server after changing the parameter.

This parameter is the RTSP port number using which IVRMP connects to the IBM WVS speechserver.

Port number configuration

Fundamentals — Speech MRCP 3 Dec 2010 31

Page 32: Fundamentals Speech MRCP

Common ports on the Media Server nodeThe following information focuses on common ports on the IBM WebSphere Voice Server 6.1(WVS 6.1). Use this information to answer any questions you have about port security orcommonly used ports on IBM WVS 6.1.

In the interest of best practices for a secure network environment, administrators should leaveopen only the ports required for system operations. You can close all ports that are not requiredfor system operations.

For more information about common ports on the Media Server node, see Table 7: Commonlyused ports on the Media Server node on page 32.

Table 7: Commonly used ports on the Media Server node

Port Use80 IBM HTTP Server Port

4554 IBM WVS Media Resource Control Protocol(MRCP) default port for RTSP

9080 Internal WAS HTTP Server

9090 Administrative Console

35000 to 60000 Ports reserved for RTP for data transmissionbetween the MRCP Client and MRCP MediaServer

Extend the time an LVR or TTS session can remain idleUse the following information as an overview for extending the time a Large-Vocabulary-Response (LVR) or Text-To-Speech (TTS) session can remain idle on the IBM WebSphereVoice Server 6.1 (WVS 6.1).

The com.ibm.voice.server.rtspbridge.sockettimeout parameter determines theamount of time an LVR or TTS session can remain idle before the server closes the connectionto the client. After the connection closes, the resource is no longer available to the applicationfor the instance of allocation during which it was idle. However, after the connection closes,the LVR or TTS resource can be used again for another instance of allocation (SETUP).

IBM WebSphere Voice Server fundamentals

32 Fundamentals — Speech MRCP 3 Dec 2010

Page 33: Fundamentals Speech MRCP

You may need to increase this value to accommodate delays in the application for resourceusage. Conditions that may require an increase in the value of this parameter follow. In thesecases, the TTS resource is idle the entire time that these operations are occurring.

• playing long pre-recorded prompts (for example, a .wav file)

• any application operation that can take a very long time, such as accessing a database

For example, if a .wav file is playing for longer than the session timeout and the LVR or TTSsessions are idle, the server closes the connection when the timer expires.

Ensure that you increase this timer so that the resource can remain idle and be available tosatisfy the needs of the application. The default is 300 seconds, which is equal to 5 minutes.

Use the following syntax to set this parameter: putParam.batcom.ibm.voice.server.rtspbridge.sockettimeout <value>This parameter is read from the IBM WVS server configuration file during system startup ofthe IBM WVS 6.1.x node. The timer is started at session start (SETUP). The parameter cannotbe modified during run time because a session (connection) is required to send any message tothe server. This parameter is configured only during system initialization. If you modify the valueof this parameter, you must restart the IBM WVS WVS 6.1.x node to effect this change.

Voice configurationUse the following information as an overview for voice configuration on the IBM WVS 6.1, andfor voice configuration, including disabling unnecessary voices and changing default voices.

Disable unnecessary installed voicesAfter you install TTS voices on the Media Server node, they consume system resources.Enable only the TTS voices that you plan to use for SPEAK requests. Disabling unnecessaryTTS voices installed on the node avoids needless consumption of system memory.

For example, if you plan to use only the female voice and not the male voice for a language,you activate the female voice. If more than one female voice is installed, activate only thefemale voice that you plan to use.

Use the scripting interface wsadmin to disable voices. For more information about setting TTSavailable voices, see the InfoCenter Web site.

In the following example, the fifth-generation U.S. English voice Lisa is installed on a MediaServer node where the third-generation voices Andrew and Allison are installed. But you nowwant to use Lisa for SPEAK requests and not Andrew or Alison. In this case, you want to remove

Voice configuration

Fundamentals — Speech MRCP 3 Dec 2010 33

Page 34: Fundamentals Speech MRCP

one male voice (Andrew) and one female voice (Allison) and activate only the female voiceLisa. Use the following commands to activate the voice Lisa:

cd C:\Program Files\WebSphere\VoiceServer\samples.

putParam.bat com.ibm.voice.server.synth.voices Lisa.

In this example Andrew, Allison, and Lisa are installed on the node but only Lisa consumessystem memory. Andrew and Allison do not consume system memory because they are nowinactive voices.

Change default voiceThe IBM WVS 6.1 allows administrators to change the default voice for speech synthesisresources. Use the following information to change the default voice.

An administrator can configure a default TTS voice for each WebSphere Voice Server.

The IBM WVS 6.1 uses the default-voice parameter to assist in the selection of a TTSvoice when a TTS SPEAK request is missing the necessary voice attributes.

You can use the gender, age, and language attributes of the default voice if a TTS SPEAKrequest does not explicitly provide these attributes.

You must select the default TTS voice name from the list of available TTS voices. The currentdefault voice appears in the WebSphere Voice Server Administrative Console.

The U.S. English default voice is Andrew. The commands in the following example change thedefault voice from Andrew to Allison:

cd C:\Program Files\WebSphere\VoiceServer\samplesputParam.bat com.ibm.voice.server.synth.default-voice Allison.

Administrative considerationsThe Administrative Console is a Web-based application that runs on the WebSphere VoiceServer 6.1 (WVS 6.1) and allows administrators to monitor speech resources. The followinginformation provides a link to the Administrative Console and briefly discuses this tool. Readthis information before you use the Administrative Console to manage resources on the IBMWVS node.

The AdminConsole is a Graphical User Interface (GUI) tool that you use to manage operationson the IBM WVS node. The URL for the Administrative Console is as follows: http://<hostname>:9090/admin/

IBM WebSphere Voice Server fundamentals

34 Fundamentals — Speech MRCP 3 Dec 2010

Page 35: Fundamentals Speech MRCP

If you cannot complete a management operation with the Administrative Console, you use thescripting interface wsadmin.

Media Server management and configurationThe following issues relate to managing the IBM WVS, where the IBM WVS 6.1 vendorsoftware is installed. Perform management and monitoring operations on the Media Servernode using the WVS Administrative Console or the scripting interface wsadmin.

The environment variable %JAVA_HOME% points to the location of the scripts and other filesused to manage the Media Server node.

Use the following command to invoke the scripting interface:

%JAVA_HOME%\..\bin\wsadmin.bat

The IBM WVS product is configured with default values for all parameters. The MRCP server isready to function after installation without additional configuration. If this configuration isconsistent with the site-specific environment, you don't need to perform an additionalconfiguration. But if you make a configuration change, you must restart the Media Server nodebefore this change takes effect.

For more information see the WebSphere Voice Server Information Center (InfoCenter) andother documentation provided by IBM.

Use the IBM Administrative Console to manage operations on the Media Server node.Alternatively, you can use a set of IBM scripts to perform the following tasks:

• start and stop Media Server

• log on to Media Server

• configuration

• obtain the current status of an IBM WVS Media Server

• voice server

• log and trace operations

• backup and restore the Media Server Restore

• restore Media Server configuration parameters to their default value

You can rename the IBM Media Server node if you received an Avaya factory installedWebSphere Voice Server machine, and you want to change the host name of that machine.

Media Server management and configuration

Fundamentals — Speech MRCP 3 Dec 2010 35

Page 36: Fundamentals Speech MRCP

Change the Media Server Node host nameThe IBM WebSphere Voice Server 6.1 allows administrators to change the host name of theMedia Server Node. The information provides background material on how to change the nodename and provides a link to this procedure. Read this information before you attempt thisprocedure.

You need to rename the IBM Media Server Node only if you received an Avaya factory installedWebSphere Voice Server machine, and you want to change the host name of that machine.

The host name is the network name for the physical machine where the node is installed. Thehost name must resolve to a physical network node on the server. Selecting a host name thatother machines can reach in your network is extremely important.

Warning:Do not use the generic localhost identifier for the Media Server Node host name.

The host name value is the value of the hostName property in WAS configuration documents.

The hostname is specified during the Avaya factory installation of the WAS and WVS software.

Warning:You can only use this procedure with a new machine. You cannot use this process if themachine has been federated into a Deployment Manager cell.

After completion, you can type hostname on the command line and have it return your<newhostname>. You can also issue the command ping <newhostname> and receive asuccessful reply. Issue these commands as follows, assuming the new host name is changedto newhostname:

> hostname > ping newhostname newhostname Pinging newhostname [9.22.93.59] with 32bytes of data: Reply from 9.22.93.59: bytes=32 time<1ms TTL=128 Reply from 9.22.93.59:bytes=32 time<1ms TTL=128 Reply from 9.22.93.59: bytes=32 time<1ms TTL=128

For a listing of the tasks involved in changing a host name, see Table 8: Tasks to completewhen changing a host name of a server. on page 37

The following table identifies the instructions you need to follow when changing the host nameof a server. Each of the six tasks must be completed in the order shown.

See IBM WebSphere Voice Server host name change on page 68 for steps on how to changea host name.

IBM WebSphere Voice Server fundamentals

36 Fundamentals — Speech MRCP 3 Dec 2010

Page 37: Fundamentals Speech MRCP

Table 8: Tasks to complete when changing a host name of a server.

TopicChange host name in WebSphere Application Server

Change host name in IBM HTTP Server

Change the host name in WebSphere Voice Server with ifix_IC51934

Update WebSphere Application Server Web Server Plug-in

Update the Administrative Console Startup Link with the New hostname

WebSphere Application Server hostname and node name

WebSphere Application Server host name and node nameThe IBM WebSphere Voice Server 6.1 allows administrators to change the host name of theWebSphere Application Server (WAS). The information provides background material on howto change the node name and provides a link to this procedure. Read this information beforeyou attempt this procedure.

The WAS host name and WAS node name are discrete fields that can have different values.

The WAS node name is displayed on the IBM Administration Console and customers use itfor administration purposes.

WAS host name is the Domain Name Service (DNS) name for the machine and is associatedwith the Internet Protocol (IP) address of the machine. See your system administrator forinformation about modifications to the DNS database and other system administrative tasks.The host name set in the DNS database and the host name set in the machine must beidentical. The host name is set on the machine using the path:

Start, Control Panel, System, Computer Name.

You cannot set the host name to different names in these two places as this causes severalknown issues. For example, when the host name is set differently, you cannot access themachine using the host name set in path Start, Control Panel, System, ComputerName because DNS settings override the host name set on the machine. Only the host nameset in the DNS database is recognized.

During installation, you enter both of these names in different fields on the same screen. In thefollowing example, the same name is entered for node name and host name. In this example,both names are oscpcv071.

For more information, see Installing IBM WVS 6.1 software on page 57.

You can change the host name of the machine after installation. However, be aware thatchanging the host name does not change the node name. The IBM Administration Console

WebSphere Application Server host name and node name

Fundamentals — Speech MRCP 3 Dec 2010 37

Page 38: Fundamentals Speech MRCP

displays the node name. The node name and host name are different fields and can have thesame or different values.

The following example shows what you can expect to see when the host name and node nameare different. In this example, the host name changes from ibmwvs154 to pvrohs154. Thenode name is ibmwvs154. It did not change with the host name change.

• After you change the host name from ibmwvs154 to pvrohs154, access the IBMAdministration Console using the address that includes the new host name. In this casethe address is: pvrohs154:9090/admin. This address appears in the Web browserAddress field.

• The node name ibmwvs154 appears in the IBM Administration Console window underthe heading Voice Server Name.

If you change the host name of the IBM WVS Media Server machine, you must edit the DNSentries and other hosts file entries on all machines that are configured with the IP address andoriginal host name and you must associate the IP address with the new host name.

Avaya does not guarantee a unique host name is assigned at the point of manufacture.administrators are responsible for ensuring that all host name and IP address issues areresolved when the IBM WVS Server machine is connected to your site-specific network . Forexample, multiple IP addresses cannot refer to the same machine.

You must ensure that the addresses are correct and that the host names associated with thoseIP addresses are valid, site-specific host names across all DNS databases and other networkrelated entities.

For more information, see Table 7: Commonly used ports on the Media Server node on page 32for a list of commonly used ports.

Backup and restoreThe network administrator is responsible for the backup and restore the IBM WebSphere VoiceServer (WVS 6.1). The following information outlines your responsibility and providesreference documentation for this task. The administrator can read this information beforeperforming this procedure on the IBM WVS 6.1.

Use the backup scripts provided by IBM to backup and restore the Voice Server configuration.These scripts reside in the following directory path: %WVS_ROOT%\samples.

For more information, see WebSphere Voice Server Information Center (InfoCenter) and otherdocumentation provided by IBM.

To ensure a secure computer network, Avaya recommends that the backup and restore of theIBM WVS be part of your overall Backup/Restore plan.

IBM WebSphere Voice Server fundamentals

38 Fundamentals — Speech MRCP 3 Dec 2010

Page 39: Fundamentals Speech MRCP

Vendor loggingThe IBM WebSphere Voice Server 6.1 (WVS 6.1) allows administrators to perform vendorlogging through the Collector tool. The following information outlines vendor logging on theIBM WVS 6.1 and discusses the Collector Tool feature. Read this information before youperform this task.

IBM WVS vendor logging occurs on the Media Server node. When you activate waveform andcall logs, and enable trace logging, files are created in the sub directory %WVS_ROOT%\ on theMedia Server node. Use the WebSphere Collector tool to collect all logs (call logs and tracelogs) from the node and to delete these logs after the collection process.

Before you use the Collector tool, create a new directory on the Media Server node for the filecollection process. When you use the Collector tool, it creates a .jar file in this directory, itcollects all call and trace logs from other directories where they were created, and it deletesthe log files from those directories. But the .jar file remains in the directory.

You must create a new directory before using the Collector tool. If you run the Collector toolwithout creating a new directory, the new .jar overwrites the existing .jar file.

You must send the .jar file to your Certified Account Manager for diagnostic evaluation.

Warning:There must be enough free disk space available before you activate logging and use theCollector tool. The IBM WVS software stops functioning when the disk space reaches 0bytes. To ensure that enough free disk space exists, you create a script that checks for diskspace. If the amount of disk space is inadequate, use the scripting interface wsadmin.batto disable waveform or other logging.

Do not delete the call and trace log files manually. Always use the Collector tool for this diskspace maintenance operation. If you do not use the Collector tool, call and trace log files remainon the disk until the free space reaches zero bytes.

The amount of time the Collector tool takes to complete the capture process and create thefiles depends on the type of logging and trace settings you configure. If you activate wave filecapture, additional time is required to complete the collection process. Be aware that usingthis tool impacts system resources, principally Central Processing Unit speed (CPU speed andmemory).

IBM recommends that after successfully completing the IBM WVS software installation andverification process, you use the WebSphere Collector Tool to obtain baseline logging aboutthe system.

The WebSphere Collector Tool is invoked by executing the following command:

%JAVA_HOME%\..\bin\collector.bat.

For more information, see the WebSphere Voice Server Information Center (InfoCenter) andother documentation provided by IBM for support information.

Vendor logging

Fundamentals — Speech MRCP 3 Dec 2010 39

Page 40: Fundamentals Speech MRCP

Avaya supports the MRCP Logging-Tag parameter.

Media Server remote supportAdministrators use the remote support feature to perform operations on the IBM Media Server.The following information focuses on the use of Telnet for remote support. Read this informationbefore you work with remote support.

You can perform most operations on the IBM Media Server with the IBM AdministrativeConsole for remote support. You can use Telnet for remote support of Media Server node in aWindows environment and Remote Desktop on the IBM Speech servers.

For more information, see the IBM WVS 6.1 Redbooks (distributed by IBM).

Grammars and lexiconsSpeech recognition uses grammars to recognize what a caller says in response to a systemprompt. Speech synthesis uses lexicons to generate correct pronunciation. The followinginformation describes the use of grammars and lexicons on the IBM WebSphere 6.1 (WVS6.1). Read this information before you work with these resources.

Typically, grammars and lexicons reside on a Web server and the Media Server must retrievethem from the Web server and save them to the Media Server node before it can satisfy arequest from the application for recognition or synthesis.

You must preload all grammars and lexicons on the Media Server node before applicationexecution. Use the IBM WVS Administrative Console to perform this task.

Preloading grammar and lexicon files provide for a faster response to application requests forrecognition or synthesis. This operation saves the time it takes to retrieve the grammar orlexicon from a Web server when responding to a request from an application.

The speech applications do not know which grammars are loaded on the Media Server. Theysimply request a grammar in preparation for a recognition. Grammar retrieval time is criticalfor successful application execution.

When the Media Server receives a request for a grammar, it checks to see if the grammar ispreloaded on the Media Server node. If the grammar is not preloaded, it retrieves the grammarfrom a Web server node. If the grammar is preloaded, the Media Server compares the Webserver version to the cached version and, if the grammars have the same date, the MediaServer uses the cached grammar. If a grammar is stale, it retrieves the new version of thegrammar from the Web server.

IBM WebSphere Voice Server fundamentals

40 Fundamentals — Speech MRCP 3 Dec 2010

Page 41: Fundamentals Speech MRCP

For more information about caching rules on the IBM WVS Media Server, see thedocumentation provided by IBM.

Grammar caching rules are configurable and you must be aware of the site-specific rules thataffect grammar caching when working with this resource.

For information about grammar cache control guidelines, see ASR Troubleshooting Tips in theIBM WVS InfoCenter.

IBM WVS toolsIBM provides a number of tools to assist you with configuring speech resources on the IBMWebSphere Voice Server 6.1, including the WebSphere Application Server NetworkDeployment Manager, the IBM Voice Tool kit, and the Voice Trace Analyzer. The followinginformation focuses on how to use these tools. Read this information before you work withthese resources.

The following IBM WVS tools are available from IBM for use on the Media Server node:

• WebSphere Application Server Network Deployment Manager: A single WebSphereApplication Server System Administrator Console to manage a network of IBM WVS 6.1.xmachines

• IBM WVS Voice Toolkit: A tool for building grammars and more. The IBM WVS VoiceToolkit contains the following tools:

- VoiceXML Application Builder

- Grammar Builder

- Grammar Compiler

- Lexicon Builder

The toolkit is only available from the IBM Toolkit Support Web site.

• Voice Trace Analyzer: A tool to examine recognition data using logs and traces collectedby the IBM WVS Collector tool. The files are collected from a single node only, and asingle set of analysis files are generated for the node. If TTS and LVR resources areinstalled on different Media Server nodes, you need to run the Voice Trace Analyzer oneach node and manually reconcile them. For more information, see Vendor logging onpage 39.

IBM WVS tools

Fundamentals — Speech MRCP 3 Dec 2010 41

Page 42: Fundamentals Speech MRCP

IBM WebSphere Voice Server fundamentals

42 Fundamentals — Speech MRCP 3 Dec 2010

Page 43: Fundamentals Speech MRCP

Chapter 6: Nuance Speech Serverfundamentals

The Nuance Speech Server (NSS) is a software product that you can use with other software andhardware telephony products to provide speech resources such as speech recognition and speechsynthesis. The following information provides an overview of the NSS in the Media Applications Server6.4 (Avaya MS 6.4) environment. Use this information when you are work with licensing, configuration, oradministrative issues.

Navigation

• Product overview on page 43

• Vendor documentation reference material on page 44

• Licensing considerations on page 45

• Configuration required on the Nuance Speech Server on page 45

• Disable real time antivirus scanning on the Nuance Speech Server on page 48

• Remote support on page 49

• User dictionary and the say as tag on page 49

• User dictionary configuration on page 51

• Lexicons on page 51

Product overviewThe Nuance Speech Server (NSS) is a Media Resource Control Protocol (MRCP)-basedspeech resource that provides speech recognition (Nuance Recognizer) and speech synthesis(RealSpeak). Read the following information about the NSS within the Media ApplicationsServer 7.0 (Avaya MS 7.0) environment before you work with the NSS.

The NSS software provides a standard protocol interface to a network speech engine(recognizer and synthesizer). The Speech Server software is a network application andcontains all system management and monitoring software you require to remotely manage aspeech recognition server.

The MRCP provides the basis for communication with the recognizer and other services. TheAvaya MRCP implementation is based on the Internet Engineering Task Force (IETF) MRCPdrafts.

Fundamentals — Speech MRCP 3 Dec 2010 43

Page 44: Fundamentals Speech MRCP

Speech Server software supports the following features:

• MRCP interface to Nuance Recognizer and RealSpeak

• a localization-ready API that handles multibyte character strings

• optimizations for VoiceXML browsers, including (secure) URL resolution, W3C grammarformat, parallel grammars, and grammar caching

• logging of waveforms (caller speech) and recognition events

• MRCP, Transport Layer Security (TLS), and Secure Real-time Transport Protocol Security(SRTP) for secure communications

The NSS supports MRCPv1 and MRCPv2. However, Avaya only supports MRCPv1.

Avaya unsupported vendor functionalityThe Nuance Speech Server (NSS) is vendor software that resides in the Avaya MSenvironment. The following information focuses on the features Avaya supports on the NSS.

Avaya supports all existing functions on the Nuance Speech Server (NSS).

Vendor documentation reference materialNuance Communications currently provides reference material for speech recognition andspeech synthesis resources. The following list contains documents for these resources thatyou can use when working with the Nuance products.

Nuance Communications provides documentation for the RealSpeak and Nuance Recognizer9.0 resources.

These documents are installed with the Nuance resource software and include the following:

• Nuance Speech Server Product Documentation

• RealSpeak 4.5 Product Documentation

• Nuance Recognizer 9.0 Product Documentation

• Nuance License Manager Licensing Enduser's Guide

This documentation reside in the following paths: Nuance Speech Server Start,Programs, NUANCE, NSS 5.0, Product Documentation RealSpeak 4.5 Start,Programs, ScanSoft, RealSpeak 4.0, Product Documentation NuanceRecognizer 9.0 Start, Programs, Nuance Recognizer 9.0, ProductDocumentation Nuance License Manager Start, Programs, Nuance LicenseManager, Licensing Enduser's Guide

Nuance Speech Server fundamentals

44 Fundamentals — Speech MRCP 3 Dec 2010

Page 45: Fundamentals Speech MRCP

The documentation also resides in the following directory paths:

C:\Program Files\Nuance\license_manager\doc\ C:\Program Files\RealSpeak 4.0\doc C:\Program Files\RealSpeak 4.0\tools C:\ProgramFiles\Recognizer\documentation\ C:\Program Files\Speech Server5.0\server\doc

Licensing considerationsThe Nuance Speech Server uses a single licensing file for speech recognition (NuanceRecognizer 9) and speech synthesis (RealSpeak 4.5).

Configuration required on the Nuance Speech ServerThe following information discusses how to configure the NSSserver.cfg file and theBaseline.xml file for Text-To-Speech (TTS) and Large-Vocabulary-Response (LVR)services on the Nuance Speech Server (NSS).

Configure the NSSserver.cfg fileThe Nuance Speech Server (NSS) is configured by parameters that reside in theNSSserver.cfg file on the NSS node.

Ensure that the NSS has enough CPU and memory to sustain the number of licensed channelsconfigured to run Nuance TTS and LVR sessions. You must configure the NSS toaccommodate the number of LVR and TTS channels that can be active simultaneously. Inaddition, you must ensure that the port on which RTSP communicates MRCP messagesbetween the NSS and the Avaya MS is set to the same value. Set the following parameters inthe NSSserver.cfg file to ensure the NSS is configured properly. These parametersbecome active at system startup of the Nuance Speech Server node.

If you modify this file, you must restart the Nuance Speech Server node for the parameters totake effect.

server.rtp.maxCountOfSessionThis value affects the number of simultaneous LVR and TTS audio streams that the NSS cansupport. Set this value to, at least, two counts for every LVR and TTS channel configured forthe NSS. For example, when the NSS is configured for 382 TTS channels and 382 LVR

Licensing considerations

Fundamentals — Speech MRCP 3 Dec 2010 45

Page 46: Fundamentals Speech MRCP

channels, the minimum required value is 1528 (382+382=764x2=1528). In this case, set thevalue to 1600, which allows for somewhat more than the 1528 minimum.

server.mrcp1.rtsp.maxCountOfSessionThis value affects the number of simultaneous LVR and TTS sessions that the NSS cansupport. Set this value to, at least, one count for every LVR and TTS channel configured forthe NSS. For example, when the NSS is configured for 382 TTS channels and 382 LVRchannels, the minimum required value is 764 (382+382=764). In this case, set the value to800, which allows for somewhat more than the 1528 minimum.

server.mrcp1.rtsp.sessionTimeoutThe server.mrcp1.rtsp.sessionTimeout parameter determines the amount of time anLVR or TTS session can remain idle before the server closes the connection to the client. Afterthe connection closes, the resource is no longer available to the application for the instanceof allocation during which it was idle. However, after the connection closes, the LVR or TTSresource can be used again for another instance of allocation (SETUP).

You may need to increase this value to accommodate delays in the application for resourceusage. Conditions that may require an increase in value of his parameter follow. In these cases,the TTS resource is idle during the time that these operations are occurring.

• playing long pre-recorded prompts (for example, a .wav file).

• any application operation that can take a very long time, such as accessing a database.

For example, if a .wav file is playing for longer than the session timeout and the LVR or TTSsessions are idle, the server closes the connection when the timer expires.

Ensure that you increase this timer so that the resource can remain idle and be available tosatisfy the needs of the application. The default is 60000 millisecs, which equals 1 minute.

server.mrcp1.transport.TCPTimeoutYou may need to decrease the default value if you are running a load test because it takestime for the channel session to cycle for reuse. This timeout affects the usage of a channelafter it is released. The NSS holds a channel for the time set for this TCPtimerout value andallows reuse or reallocation of this channel only after this timeout value. In order to allowimmediate use, set this value appropriately.

server.mrcp1.transport.portThis value indicates the port on which RSTP communicates between the NSS server and theAvaya MS node. The default value is 4900.

For additional information about the Nuance Speech Server, follow the menu path: Start,Programs, NUANCE, documentation, NSS 5.0, Product Documentation.

Nuance Speech Server fundamentals

46 Fundamentals — Speech MRCP 3 Dec 2010

Page 47: Fundamentals Speech MRCP

Configure the NSSserver.cfg file for co-residencyTo configure the Nuance Speech Server for co-residency with the Avaya MS, the port numbersfor the following parameters must be changed to avoid port conflicts on the Avaya MS. Ensurethat no other software uses the new port numbers, otherwise the system may not functionproperly.

Avaya does not support co-residency of the Nuance Core Services (Nuance's watcherdaemon, file transfer service and resource manager) with Avaya MS. As the operating systemis a hardened one, any TCP ports specified below need to be opened up on the node. However,it is best not to enable the mrcp2 ports, as this is not supported by Avaya. The port numbersin the configuration file need to be changed in order for proper functioning of Avaya and Nuancesoftware as it is not possible to disable the mrcpv2 sip ports in Nuance's software.

server.mrcp2.sip.transport.tcp.portChange the default value of 5060 to the recommended value of 5070.

server.mrcp2.sip.transport.udp.portChange the default value of 5060 to the recommended value of 5070.

server.mrcp2.sip.transport.tls.portChange the default value of 5061 to the recommended value of 5071.

server.mrcp2.transport.tcp.portChange the default value of 6075 to the recommended value of 6095.

server.mrcp2.transport.tls.portChange the default value of 6076 to the recommended value of 6096.

server.session.ossweb.portChange the default value of 80 to the recommended value of 85.

Important:Change the port number for the Apache Web Server (in the httpd.conf file) to the samenumber used for server.session.ossweb.port.

server.mrcp1.transport.portThe default port number of 4900 can remain the same, however, if it is changed then the portnumber in Element Manager (under MRCP Servers) must also be changed.

Configure the NSSserver.cfg file for co-residency

Fundamentals — Speech MRCP 3 Dec 2010 47

Page 48: Fundamentals Speech MRCP

Configure the Baseline.xml fileThe Baseline.xml file contains Nuance Recognizer 9.x parameters and resides in thedirectory path %SWISRSDK%\config (C:\Program Files\Nuance\Recognizer).These parameters become active at system startup of the Nuance Speech Server node. If youmodify this file, you must restart the Nuance Speech Server node for the parameters to takeeffect.

The Nuance Extra Nbest Keys feature is used to add additional grammar keys to the XMLresult. This feature must be enabled or disabled on both the Nuance Speech Server and theVoiceXML interpreter to ensure proper operation. To enable this feature on the Nuance SpeechServer, uncomment the swirec_extra_nbest_keys parameter below in theBaseline.xml file. To disable this feature, comment out the entries. To enable or disablethis feature on the VoiceXML interpreter, see the Avaya Media Server Configuration VoiceXMLand CCXML Application Programming, NN44471-501.

The Nuance Speech Server must not be configured with the parameterswirec_extra_nbest_keys. The VoiceXML interpreter cannot parse the speech results ifthis parameter is active. If the Baseline.xml file contains an entry forswirec_extra_nbest_keys, comment out the entry as in the following example.

<!-- Add a Nuance grammar key to the XML result.--> <!-- <paramname="swirec_extra_nbest_keys"> <value>SWI_meaning</value><value>SWI_literal</value> <value>SWI_grammarName</value> </param>-->To modify this file, first, create a backup copy of the file. Edit the file using the WordPad editorand not the Notepad editor, as Notepad corrupts the file. Some editors either do not format theXML code correctly or insert characters that have caused problems that prevent the file fromworking properly after the change. Avaya recommends to use WordPad to edit this file.

For additional information about Nuance Recognizer 9.x, refer to the vendor documentationthat is available from the menu path:

Start, Programs, Nuance Recognizer 9.0, Product Documentation.

Disable real time antivirus scanning on the Nuance SpeechServer

If antivirus software is installed on the Nuance Speech Server, do not enable scanning duringreal time operations in a production environment.

Nuance Speech Server fundamentals

48 Fundamentals — Speech MRCP 3 Dec 2010

Page 49: Fundamentals Speech MRCP

Real time scanning has a negative impact on system operation and performance.

• At best, execute antivirus scanning during maintenance periods, when Nuance SpeechServer processes are shut down.

• At minimum, limit antivirus scanning to the hours when call traffic is lowest and excludeall folders that contain Nuance software from the scan.

In the interest of overall network virus protection, it is best to configure the Avaya solution ona separate subnet to isolate it from unnecessary traffic. You can take additional measures tosecure the network from unnecessary traffic using firewalls that allow restricted (Virtual PrivateNetwork) VPN or remote access. You can also limit communications to only the subnets, IPaddresses, or ports that are required for proper operations.

The best way to ensure that a system can cope with the additional CPU, disk, and networkbandwith that virus scanning can consume, is to consider virus scanning, initially, whenperforming system provisioning and sizing.

Both Avaya and Nuance Communications, Inc. recommend this approach to virus security.

Remote supportNSS is equipped for remote support. Use remote support software such as Windows RemoteDesktop to connect with the NSS node.

User dictionary and the say as tagA user dictionary is a collection of word pronunciations used by RealSpeak when it convertstext to the spoken word. The following information focuses on how user dictionaries functionon the Nuance Speech Server (NSS), including how to create and edit a user dictionary, how toconfigure a user dictionary, and how to use Say-As tags to maximize speech synthesis on theNSS.

Use the user dictionary to specify special pronunciations for words or character strings and tosubstitute the original text strings with

• words that you want spoken to the caller

• words or character strings that are not in the existing sound bank

Each dictionary entry specifies a Source Text string and a corresponding Destination Textstring. The dictionary substitutes the Destination Text string for the Source Text string. TheDestination Text string is synthesized and spoken to the caller.

Remote support

Fundamentals — Speech MRCP 3 Dec 2010 49

Page 50: Fundamentals Speech MRCP

For example, if the Source Text string is nt and you set the Destination Text to Avaya, thesynthesizer speaks the name Avaya to the caller each time the text to be synthesized containsthe text string nt.

For more information about upgrading a dictionary from a previous release to RealSpeak 4.5,see the documentation provided by Nuance Communications, Inc.

User dictionaries are language-specific. RealSpeak provides examples of user dictionaries inthe following directory: %SSFTTTSSDK%\api\demos\data.

RealSpeak provides a user dictionary Editor tool that you can use to create and edit userdictionaries. You run this tool in one of two ways:

• type rsude at an MSDOS command line prompt

• follow the path C:\Program Files\ScanSoft\RealSpeak 4.0\speech\components\common\ and select Launch rsude.exe.

Typically, the Speech Server node is not configured with a speaker. The user dictionary Editortool checks for the speaker when it starts and if the Editor Tool does not locate a speaker, anError window opens with an Initialization Error message.

Click OK to open the user dictionary tool.

Substitution is not case-sensitive. The replacement string can be orthographic (correctspelling) or phonetic (spelled to represent the sound of the word) text.

Dictionaries can be in ASCII text and the contents are organized into sections with the labels[Header], [SubHeader] and [Data].

Two important points that you must be aware of:

• Each dictionary begins with a [Header] label that describes dictionary properties, suchas the language

• Each [Header] label is associated with one or more [SubHeader] and [Data] labelpairs

- The [SubHeader] label describes properties associated with the paired [Data]label.

- The [Data] label contains the source and destination (replacement) strings.

For more information about the dictionary format, parameters, and options, go to the NuanceCommunications Web site: http://www.nuance.com/.

Nuance Speech Server fundamentals

50 Fundamentals — Speech MRCP 3 Dec 2010

Page 51: Fundamentals Speech MRCP

User dictionary configurationTo configure the user dictionary, you must add the entry in the C:\Program Files\ScanSoft\RealSpeak 4.0\config. You must confirm that the URL is reachable. Forexample:

<default_dictionaries> <dictionary priority="1000"> http://47.185.23.32:8080/us_english_sample.dct </dictionary> </default_dictionaries>After you have configured the user dictionary you must restart the NSS service for the changesto take effect.

LexiconsThe following information focuses on lexicons on the Nuance Speech Server (NSS). Use thisinformation when you load a user dictionary through the lexicon element.

Nuance supports loading of SSFT user dictionaries through the SSML lexicon element. Thevalue for the Uniform Resource Identifier (URI) attribute must be a valid URL to an SSFT userdictionary. The dictionary can be in one of the two supported SSFT formats; textual (forexample, *.dct or *.tdc) or binary (for example, *.dcb *.bdc). The TYPE attribute is optional.Valid values are

• application/edct-bin-dictionary

• application/edct-text-dictionary

When you use an HTTP server, you must add two entries to your MIME table to associate thedictionary extension with the MIME type. For local file access, the following file extensions arecorrectly mapped on the box: .dct, .tdc, .bdc, and .dcb.

Example <lexicon> to use within a VXML application:

<prompt> <lexicon uri="http://47.185.23.32:8080/us_english_sample.dct" type="application/edct-text-dictionary"/>DLL , we should not hear D L L </prompt>For more information on lexicons, see the SSML documentation provided by NuanceCommunications.

User dictionary configuration

Fundamentals — Speech MRCP 3 Dec 2010 51

Page 52: Fundamentals Speech MRCP

Say-as tagsSpeech Synthesis Markup Language (SSML) say-as tags are designed to assist with thegeneration of speech synthesis in the Avaya MS environment by providing for the pronunciationof e-mail addresses. The following information focuses on support for the say-as tags.

While the W3C SSML 1.0 Recommendation specifies the sayas element and its semanticswith the interpret-as, format, and detail attributes, it does not define a specific sayastypes. Standardized values for the sayas attributes are published in a W3C Group Note.

Nuance is involved in the W3C and is committed to supporting these guidelines.

RealSpeak supports both the sayas syntax of the April 2002 SSML which specifies the typeattribute, the December 2002 syntax, and any recent versions.

For a full listing of supported sayas tags see the RealSpeak 4.0 documentation found at C:\Program Files\ScanSoft\RealSpeak 4.0\doc.

Nuance Speech Server fundamentals

52 Fundamentals — Speech MRCP 3 Dec 2010

Page 53: Fundamentals Speech MRCP

Chapter 7: IBM WebSphere Voice Serverinstallation and uninstallation

The following information provides the steps to install and uninstall the IBM WebSphere Voice Server 6.1(WVS 6.1). Use it to answer any questions you have about the installation of this software including how tochange a host name.

Navigation

• IBM WebSphere Voice Server installation on page 53• IBM WebSphere Voice Server uninstallation on page 64• IBM WebSphere Voice Server host name change on page 68

IBM WebSphere Voice Server installationThe following information provides instructions about installing the IBM WebSphere 6.1 VoiceServer (WVS 6.1) in the Media Applications Server 7.0 (Avaya MS 7.0) environment.

These instructions are based on the IBM software made available to Avaya at publication.While the source of some of this information is the Preload Guide provided to Avaya by IBMand the IBM InfoCenter Web site, these instructions have been documented by Avaya and notIBM. For more information about installing or uninstalling IBM software, see the appropriateIBM documentation.

This following information addresses the installation of IBM software on the Media Server nodeinstalled with Windows 2008.

Prerequisites

Avaya recommends the following requirement guidelines for the IBM WVS 6.1.1 Media Servernode. You must meet these prerequisites before installing software on the IBM WVS 6.1 MediaServer node.

First, verify that the installation server meets the following specifications:

• a Windows 2008 and Service Pack 2 is being used• an IBM eServer X Series 336 Dual Processor is being used• Intel x86 1 GHz or equivalent processor.• a minimum of 2GB RAM is available for the recognition resource or the synthesis resource

for all languages except fifth-generation languages• a minimum of 4GB RAM is available for the synthesis resource with fifth-generation voices

Fundamentals — Speech MRCP 3 Dec 2010 53

Page 54: Fundamentals Speech MRCP

• 4 GB of free disk space is available

Important:The amount of disk space listed above allows for one male and one 5th GenerationVoice; however, disk space requirements vary depending on the voices used. For moreinformation, see the table "Text-to-Speech (TTS) server disk space requirements byvoice" below.

• a static IP address is assigned to the server• no antivirus software is installed on the server since this retards server performance• a firewall is not in use• a DVD player is installed since the IBM WVS 6.1 software is distributed on 3 DVDs created

by Avaya from which you install the softwareSecond, log on to the Media Server node as administrator to install and uninstall software. Thechanges you make to the administrator password after installation do not affect the behaviorof the IBM software.

Third, you must disable the Apache or HTTP servers that are installed on the node. The IBMWVS installation fails if you do not disable these servers.

To disable a server, stop the service associated with the server. To stop a service, follow thepath Start > Settings > Control Panel > Administrative Tools > Services. Stop the Apache orHTTP service that is running on the node.

Fourth, you must check to see if the contents of the Avaya installation DVD are complete. TheIBM WVS 6.1.1 software resides on DVDs distributed by Avaya and the DVDs were createdby Avaya for the purpose of installing the IBM software. IBM created the software, and theDVDs do not contain Avaya software.

You must use three DVDs (DVD1, DVD2, DVD3) to install the software. DVD1 and DVD2contain the IBM base, fixpack, and iFix software and a selection of different languages. DVD3contains fifth-generation voices. The folders are in alphabetical order when you display them.

For more information about the contents of the installation DVDs, see Table 9: Avayainstallation DVD on page 54.

Table 9: Avaya installation DVD

DVD1 content DVD2 content DVD3 contentwas was Julie

was51fp1 was51fp1 Kate

was513base was513base Lisa

fixpack3WVS fixpack3WVS

ifixes ifixes

asrenau asrdede

asrengb asrjajp

asrenus asrzhcn

IBM WebSphere Voice Server installation and uninstallation

54 Fundamentals — Speech MRCP 3 Dec 2010

Page 55: Fundamentals Speech MRCP

DVD1 content DVD2 content DVD3 contentasresmx ttsdede

asrfrca ttsrjajp

ttsengb ttszhcn

ttsenus

ttsesmx

ttsefrca

Table 10: Text-to-Speech (TTS) server disk space requirements by voice

Language WebSphere VoiceServerMale Female/2nd 5th Generation

French Canadian Pierre - 180M Louise - 296M

German Dieter - 289M Erika - 342M Dieter5 - 1.6G Brigit -1.6G

Spanish Rafael - 187M Sofia - 342M

UK English David - 288M Dawn -280M

Sally - 268M Kate - 616M

US English Andrew - 278M Tyler- 326M

Allison - 322M Julie - 1.4G Lisa -1.4G Michael - 1.6G

To install, choose between DVD1 and DVD2 depending upon the language you want to install.For example, to install U.S. English (asrenus) and UK English (asrgb) use DVD1. Alternatively,to install Japanese (asrjajp) and Chinese (asrzhcn) use DVD2.

If you want to install languages that reside on different DVDs, you begin the installation withDVD1 or DVD2, complete the entire installation for the IBM software and the languages youwant to install. After you finish the install with that DVD, use the other DVD to add a language tothe Media Server. For example, if you want to install U.S. English (asrenus) and German(asrdede) begin and complete the install with DVD1 installing U.S. English ((asrenus). Nextyou add the German (asrdede) language using DVD2.

If you want to install a fifth-generation language, use DVD3 after completing the install withDVD1 or DVD2.

The DVD contains the following software and is either required or optional:

The IBM WVS 6.1 software resides on DVDs distributed by Avaya. Execute the following stepsbefore installation:

• Create the IBM_WVS_6.1 directory on the local drive.• Copy the contents of the DVDs onto the local drive in the directory created in the previous

step.

IBM WebSphere Voice Server installation

Fundamentals — Speech MRCP 3 Dec 2010 55

Page 56: Fundamentals Speech MRCP

• The resulting directory provides the instructions to follow as shown below:

IBM_WVS_6.1

- asrLanguages - rhino15R3 - wvsbase - wvsSupplemental -WVS_TTS_CONCAT_ENUS_6.1.1 - wvs_tts_concat_engb_6.1.1

Important:The IBM toolkit is not distributed as part of the Avaya software distribution.

• IBM WVS 6.1 base software and related components (required).• US and UK English language for Text to Speech (at least one is required).• US and UK English language for Speech Recognition (at least one is required).• 5th Generation Voice Julie (optional).• Miscellaneous scripts (optional).

Finally, you must launch the InstallShield Wizard for IBM WVS 6.1.

The InstallShield Wizard allows you menu access to four options: IBM WVS Product Overview,Getting Started Guide, Install WebSphere Voice Server, and Install Administrative ConsoleExtensions. Use this InstallShield to obtain access to these options.

The Install WebSphere Voice Server option installs the software required for IBM WVS 6.1.1speech processing including the IBM WVS 6.1.1 base software, Text-To-Speech (TTS)languages, and Automatic Speech Recognition (ASR) languages.

The wvs513base folder contains the file Launchpad.exe and this file starts the InstallShieldwizard. You must log on to the Media Server node as Administrator to install and uninstallsoftware.

Double-click the wvs513base folder on the DVD ROM drive in Windows Explorer to exposethe contents of this folder. The DVD ROM drive in this sample installation is M:. The fileLaunchpad.exe resides in this folder.

Double-click Launchpad.exe to begin the InstallShield wizard.

Before installing the WebSphere Voice Server, you must install the WebSphere ApplicationServer and IBM HTTP Server.

• %SWROOT% represents the directory where the IBM WVS 6.1 software has been copiedfrom the DVDs.

• %SWROOT% corresponds to LocalDrive: \ IBM_WVS_6.1 for the instructions providedbelow.

Warning:During the installation, the installer checks the system requirements and proceeds with theinstallation if these requirements are satisfied. The installer does not check to determine ifsufficient RAM is available or the available capacity, such as whether the system is capableof running 50 engines or 500 engines.

IBM WebSphere Voice Server installation and uninstallation

56 Fundamentals — Speech MRCP 3 Dec 2010

Page 57: Fundamentals Speech MRCP

Installation timeUse the following information for installing the IBM WVS 6.1 in Avaya MS environment.

The Avaya implementation of speech-based transaction processing with IBM 6.1.x resources,requires you to install software distributed by IBM on each Server node and software distributedby Avaya on each Speech Server node.

On each IBM Media Server node, allow approximately one hour to install the software. Thisincludes installing the base IBM WVS 6.1 software, one language Automatic SpeechRecognition (ASR) and one language for TTS, FixPacks, and Interim fixes. The amount of timedepends on several factors including:

• the number of languages you install

• whether you install a fifth-generation voice or a third-generation language

• the CPU power of the Media Server node

• whether you install from a remote site (speed of network connection) or from a consoleconnected to the Media Server node

Installing IBM WVS 6.1 softwareInstall the IBM WVS 6.1 software for Automatic Speech Recognition (ASR) and Text-To-Speech (TTS) speech processing by following the procedures below. The installation processinstalls the base software, your choice of a language for the Large Vocabulary Recognition(LVR) resource and your choice of a voice (other than fifth-generation voices) for the TTSresource. You must install this software on the IBM Media Server node. You install the fifth-generation language Julie, as part of the post installation process.

Navigation

• Installing the Application Server Network Deployment 6.1 on page 57• Installing the HTTP Server 6.1 on page 58• Installing the WebSphere Voice Server on page 59•

Installing the Application Server Network Deployment 6.1Use the following procedure to install the Application Server Network Deployment 6.1.

Installation time

Fundamentals — Speech MRCP 3 Dec 2010 57

Page 58: Fundamentals Speech MRCP

1. Navigate to %SWROOT%\wvsbase\WAS\WAS on the local drive and double-clickinstall.exe.

2. At the Welcome to the InstallShield Wizard window, press Next.

3. At the License Panel window, read the License Agreement. Select I accept theterms in the license agreement and press Next.

4. After the prerequisites panel is completed, click Next.

5. Select the Installation Directory. Click Next.

6. You are prompted to choose the directory into which the WebSphere ApplicationServer will be installed. The default location is C:\Program Files\IBMWebSphere\AppServer. Click Next.

7. On the WebSphere Application Server Environments panel, select ApplicationServer. Click Next.

8. Select Enable administrative security on the Enable Administrative Securitypanel. Fill in the user name and password then click Next.

9. After installing the WebSphere Application Server, read the summary panel, clickNext to begin the installation.

10. Installation begins and progresses.

11. The product is then registered.

12. When installation is complete, click Finish.

13. WebSphere Application window pops up. Click Installation verification under FirstSteps to ensure that installation is successful.

Installing the HTTP Server 6.1Use the following procedure to install the HTTP Server 6.1.

1. Navigate to %SWROOT%\wvsSupplemental\IHS on the local drive and double-click install.exe.

2. Click Next on the Welcome panel.

3. Accept the license agreement and click Next.

4. After the prerequisites panel completes, click Next.

5. Accept the default port values 80 and 8008, click Next.

IBM WebSphere Voice Server installation and uninstallation

58 Fundamentals — Speech MRCP 3 Dec 2010

Page 59: Fundamentals Speech MRCP

6. On the HTTP Administration Server Authentication panel, select the following:

• Run IBM HTTP Server as a Windows Service

• Run IBM HTTP Administration as a Windows Service

• Logon as a local system account

Click Next.

7. Uncheck Create a user ID for IBM HTTP Administration Server authenticationand click Next.

8. Accept the defaults on the next panel to set up the WebSphere plugins and clickNext.

9. Read the summary panel and click Next to begin the installation.

10. Installation begins and then the product is registered.

11. Click Finish when installation is done.

Installing the WebSphere Voice ServerUse the following procedure to install the WebSphere Voice Server.

Important:You can install one or more languages during the initial install procedure provided you havethese languages available on the local disk as shown in the "Contents of the DVDs". If youwish to install additional languages, after the installation finishes, run the installation againand select another language or languages that have been copied to the local disk.

1. Open a command window, cd to %SWROOT%\wvsbase and run launchpad.bat.

2. Click Install WebSphere Voice Server to launch the installation wizard.

3. The system displays an Install for WebSphere Voice Server program has beenlaunched message, click OK.

4. The LaunchPad Welcome panel displays, click Next.

5. Accept the license agreement and click Next.

6. Select the features you want to install, including the WebSphere Voice Server TTSand/or ASR capabilities that you require. Click Next to continue.

7. Select Admin Console-Local, US English ASR and US English TTS. Click Next.

8. The installation panel for Rhino version 1.5R3 is displayed. Rhino is an open-source implementation of JavaScript written in Java.

Installing the WebSphere Voice Server

Fundamentals — Speech MRCP 3 Dec 2010 59

Page 60: Fundamentals Speech MRCP

Important:A later version of Rhino cannot be installed. This exact version must be installedfor IBM WVS 6.1 to work. Although this can be downloaded from the website: http://www.mozilla.org/rhino/download.html, it is provided as part of the Avayadistribution DVDs.

9. At the installation panel, click Choose and choose the location of the js. jar file andclick Next.

10. Choose the destination directory for WebSphere Voice Server. Click Next tocontinue.

11. Review the product information and click Install.

12. Enter the user name and password when login at the target server window pops up.

13. You are then prompted to provide various DVD locations for each component. Enterthe path and the file name for ASR, then click Next.

14. Enter the path and file name for TTS, then click Next.Example: %SWROOT%\WVS_TTS_CONTACT_ENUS_6.1.1\ttsenus\setup.exe

Important:The TTS installation is lengthy. Do not cancel out of it and do not click the Backbutton. However, if the install is cancelled when prompted for the installation files,the installation continues. The cancelled feature is bypassed but is still added tothe product registry as if it was installed. To later install the cancelled feature, firstrun the uninstaller to remove the cancelled feature from the product registry, thenselect the feature again for installation.

15. A summary panel of the installation with instructions to reboot appears. Click Doneto close the installation wizard.

16. Restart the system and run the firststeps command.

Verifying the software installationVerify software installation on the IBM WebSphere Voice Server 6.1 (WVS 6.1 by following thisprocedure. You must run the FirstSteps for the Application Server and Voice Server to ensurethat the installations were successful. Run FirstSteps after installing the IBM WVS 6.1 basesoftware and components, FixPacks, iFixes, or fifth-generation voices and after uninstallingsoftware to confirm that the system is functioning.

This verification process provides feed back about the success of the software install oruninstall. It confirms success by performing MRCP requests for LVR and TTS operations.

Prerequisites

IBM WebSphere Voice Server installation and uninstallation

60 Fundamentals — Speech MRCP 3 Dec 2010

Page 61: Fundamentals Speech MRCP

After installing the base software, restart the Media Server by following the path Start, ShutDown, and selecting Restart.

1. Open an MS DOS window and change the directory to D:\Program Files\WebSphere\VoiceServer\firststeps.

2. At the prompt, type firststeps.bat file.You can choose the FirstSteps program by selecting Start, All Programs, IBMWebSphere, Voice Server v6.1.1, FirstSteps.

3. Select Verify Installation at the Verification window.

4. After the verification completes, examine the entries in the log window to make sureall tests have passed.If the verification message indicates a failure, contact your Authorized AvayaAccount Representative.

Posting the installation procedurePost the installation procedure by following this procedure.

1. Open an MS DOS command window.

2. Change directory by entering the following %WVS_ROOT%\..\..\AppServer\bin.

3. Ensure that the server is running by opening the IBM WVS Administrative Consoleto verify the state.

4. Run setupCmdLine.bat to setup the necessary environmental variables.

5. Restart the system by selecting Start, Shutdown.

6. Start the WAS server and verify installation using Firststeps.

Installing additional third-generation languagesInstall additional third-generation language on the IBM WebSphere Voice Server 6.1 (WVS6.1) in the Media Applications Server environment.

Posting the installation procedure

Fundamentals — Speech MRCP 3 Dec 2010 61

Page 62: Fundamentals Speech MRCP

Warning:Before you install multiple languages, uninstall a language, install or add a new languageover an existing IBM WVS base, you must refer to the IBM WVS Web site for additionalinformation.

1. Navigate to the appropriate languages folder and double-click the setup.exe file tobegin the installation program.

2. The Welcome to the InstallShield window opens. Click Next.

3. Click OK to begin the install.

4. Click Next at the InstallShield Welcome window.

5. At the License Panel window, you read the License Agreement. Select I accept theterms in the license agreement and click Next.

6. Choose the WebSphere Voice Server TTS and ASR capabilities from the languagepanel for WebSphere Voice Server. Click Next to continue.

7. Click Next at the summary panel.

8. A Progress Bar displays the status of the installation.

9. Click Finish to close the installation wizard.

10. Restart the server.

11. After the system restarts, log on and follow the steps to run firststeps.bat to verifythe installation.

Installing fifth-generation TTS voicesInstall fifth-generation Text-To-Speech (TTS) voices on the IBM WebSphere Voice Server 6.1(WVS 6.1) by following this procedure.

You have the option to install and use fifth-generation voices for the TTS resource. Fifth-generation voices produce higher quality speech output. However, they require more systemresources (CPU and RAM) than the voices installed with IBM WVS 6.1 software.

IBM WebSphere Voice Server installation and uninstallation

62 Fundamentals — Speech MRCP 3 Dec 2010

Page 63: Fundamentals Speech MRCP

Prerequisites

• Fifth-generation voices require at least 4 Gb of memory on the IBM Media Server nodeto function. You must ensure that the Media Server meets this requirement or theinstallation fails.

• The 5th Generation TTS voices are shipped on the same install image as the base TTSinstall. The install image is on DVDs issued by Avaya with the IBM WVS 6.1 software.The voice is a product of IBM.

• Currently for US English, WVS installs one 2nd Generation voice (Andrew) and one 5thGeneration voice (Lisa) by default. You can choose the voice Lisa when configuring theMedia Server for TTS.

Warning:Be aware that if you uninstall languages, both the languages and the TTS engine areuninstalled.

1. Navigate to the %SWROOT%\WVS_TTS_CONCAT_ENUS_6.1.1\language\voicedirectory. Double-click setup.exe

2. The Welcome to the InstallShield window appears. Click Next.

3. Enter the directory where the voice will reside. Do not change the directory from thesetting in this window. The voice must install into this directory. The directory ischosen by the installer based on the setting of the environment variable%WVS_ROOT% which points to the base IBM WVS directory of C:\Program Files\IBM\WebSphere\VoiceServer. Click Next.

4. A summary panel displays the voice to be installed. Click Next to start theinstallation process.

5. A progress bar displays the installation status.

6. After the installation, the summary information window indicates if the install issuccessful. Click Finish.

7. Reboot the system by selecting Start, Shut Down.

8. After the system restarts, log on and follow the steps to run firststeps.bat to verifythe installation.

Checking the fifth-generation TTS voice installCheck the fifth-generation Text-To-Speech (TTS) voice install on the IBM WebSphere VoiceServer 6.1 (WVS 6.1) in the Media Applications Server 7.0 (Avaya MS 7.0) environment byfollowing this procedure.

Checking the fifth-generation TTS voice install

Fundamentals — Speech MRCP 3 Dec 2010 63

Page 64: Fundamentals Speech MRCP

1. Open the WebSphere Administrative Console, and then click Voice Servers in theleft frame.

2. A Voice Server window appears. Click the <node name> server1 link, and thenselect the Configuration tab. Information about the voice Julie appears in the VoicesInstalled window.

3. When the Media Server node shuts down, the following messages appear in theMS DOS command window:D:\Program Files\WebSphere\AppServer\bin>startServer server1 ADMU0116I:Tool information is being logged in file D:\Program Files\WebSphere\AppServer\logs\server1\startServer.log ADMU3100I: Reading configuration for server:server1 ADMU3200I: Server launched. Waiting for initialization status. ADMU3000I:Server server1 open for e-business; process id is 2828

IBM WebSphere Voice Server installation documentationIBM provides additional documentation on installing the IBM WebSphere Voice Server. Formore information, see the following list of supplementary documentation.

WebSphere Voice Server Product Overview: http://www.ibm.com.

IBM WebSphere Voice Server Information Center: http://www.ibm.com.

IBM WebSphere Voice Server Redbooks and Redpapers: http://www.ibm.com.

The IBM WVS Information Center (InfoCenter) allows you comprehensive information aboutall aspects of working with IBM WVS and is a central reference resource. See the InfoCenterfor information about installing and uninstallling the IBM WVS Media Server. You can connect tothe InfoCenter at http://www.ibm.com.

IBM WebSphere Voice Server uninstallationThe following information provides instructions to uninstall software on the IBM WebSphereVoice Server 6.1 (WVS 6.1) in the Media Applications Server environment.

Navigation

• Uninstalling IBM WVS 6.1 software on page 65• Uninstalling additional voices on page 65• Uninstalling Web server plug-ins for WebSphere Application Server on page 67

IBM WebSphere Voice Server installation and uninstallation

64 Fundamentals — Speech MRCP 3 Dec 2010

Page 65: Fundamentals Speech MRCP

• Uninstalling IBM HTTP Server on page 67• Uninstalling Application Server on page 68

Uninstalling IBM WVS 6.1 softwareUse the following procedure to uninstall the IBM WebSphere Voice Server 6.1 from the MediaServer node, which includes the base software and components.

Important:Before uninstalling, verify that you have no open Web browsers that are accessing theadministrative console. Also if you have saved any scripts in the installation paths ofWebSphere Voice Server or WebSphere Application Server, they may be deleted when youuninstall the products. Finally, if you installed a fix pack for WebSphere Voice Server,uninstall the fix pack before uninstalling WebSphere Voice Server.

To install or uninstall using a Windows® platform, log in as a user with Administratorprivileges.

1. Login to the IBM Media Server node as Administrator.

2. Stop the WVS server through the administrator console tool.

3. Navigate to the root directory %WVS_ROOT%\uninstall and start the uninstallationwizard by double clicking uninstall.exe.

4. Click Next.

5. Select Complete Uninstall then click Next.

6. After the uninstall is done, click Done then manually remove all the directoriesshown in the window.

7. Remove Voice Server directory under C:\Program Files\IBM\WebSphere.

Uninstalling additional voicesThe following describes the steps to take if you want to uninstall additional voices on an existinginstallation.

Note: To uninstall a voice, you need to run the uninstaller for the specific voice.

Uninstalling IBM WVS 6.1 software

Fundamentals — Speech MRCP 3 Dec 2010 65

Page 66: Fundamentals Speech MRCP

1. Prior to uninstalling a voice, change the default voice to one of the standard voices.

2. Stop WebSphere Application Server.

3. You need to run the uninstaller in the specific directory for the voice listed below.

4. Go to the specific directory for the voice you want to uninstall listed below, doubleclick on uninstaller.exe.

5. In the Welcome panel, click Next to continue.

6. A summary panel displays your uninstallation selections. Click Next to start theuninstallation process.

7. A summary panel of the uninstallation appears. Click Finish to close the installationwizard.

8. Reboot the system by selecting Start > Shut Down > Restart.

ResultTable 11: Voice Directories

Voice DirectoryAndrew (USEnglish)

%WVS_ROOT%\IBM\uninstallers/_uninst_ibmtts_En_US_Voice_1at08K Hz/

Allison (US English) %WVS_ROOT%\IBM\uninstallers/_uninst_ibmtts_En_US_Voice_2at08K Hz/

Tyler (US English) %WVS_ROOT%\IBM\uninstallers/_uninst_ibmtts_En_US_Voice_6at08K Hz/

Julie (US English) %WVS_ROOT%\IBM\uninstallers/_uninst_ibmtts_En_US_Voice_8at08K Hz/

Lisa (US English) %WVS_ROOT%\IBM\uninstallers/_uninst_ibmtts_En_US_Voice_7at08K Hz/

David (UK English) %WVS_ROOT%\IBM\uninstallers/_uninst_ibmtts_En_GB_Voice_1at08K Hz/

Dawn (UK English) %WVS_ROOT%\IBM\uninstallers/_uninst_ibmtts_En_GB_Voice_6at08K Hz/

Sally (UK English) %WVS_ROOT%\IBM\uninstallers/_uninst_ibmtts_En_GB_Voice_2at08K Hz/

Kate (UK English) %WVS_ROOT%\IBM\uninstallers/_uninst_ibmtts_En_GB_Voice_7at08K Hz/

IBM WebSphere Voice Server installation and uninstallation

66 Fundamentals — Speech MRCP 3 Dec 2010

Page 67: Fundamentals Speech MRCP

Uninstalling Web server plug-ins for WebSphereApplication Server

Use the following procedure to uninstall the Web server plug-ins for WebSphere ApplicationServer.

1. Stop IBM HTTP Server 6.1 and IBM HTTP Administration 6.1 through the ControlPanel, Administrative Tools, Services tool. This allows the uninstaller programto change the Web server configuration.

2. Navigate to C:\Program Files\IBM\HTTPServer\Plugins\uninstall and double clickon uninstall.exe.

3. Click Next on the Welcome screen.

4. Click Next on the confirmation screen to begin uninstalling the plug-ins.

5. Click Finish to close the Uninstaller wizard.

Uninstalling IBM HTTP ServerUse the following procedure to uninstall the IBM HTTP Server.

1. Stop the IBM HTTP Server 6.1 and IBM HTTP Administration 6.1 through theStart >Control Panel > Administrative Tools > Services tool. This allows the uninstallerprogram to change the Web server configuration.

2. Navigate to the directory C:\Program Files\IBM\HTTPServer\uninstall and doubleclick on uninstall.exe.

3. Click Next on the Welcome screen.

4. Click Finish.

5. Remove the directory HTTPServer under C:\Program Files\IBM.

Uninstalling Web server plug-ins for WebSphere Application Server

Fundamentals — Speech MRCP 3 Dec 2010 67

Page 68: Fundamentals Speech MRCP

Uninstalling Application ServerUse the following procedure to uninstall the Application Server.

1. Navigate to C:\Program Files\IBM\WebSphere\AppServer\uninstall and doubleclick on uninstall.exe.

2. Click Next on the welcome screen.

3. Click Next.

4. Click Next.

5. Click Finish.

6. Remove directory AppServer under C:\Program Files\IBM\WebSphere.

7. Remove directory WebShpere under C:\Program Files\IBM.

8. Change to the directory C:\WINDOWS.

9. Remove file vpd.properties .

10. Reboot using Start > Shut Down > Restart.

11. At this point, the IBM software is uninstalled and the Media Server node is in a cleanstate.

IBM WebSphere Voice Server host name changeThe following information provides instructions about changing the host name on the IBMWebSphere Voice Server 6.1. You must complete each task in the order shown. You need torename the IBM Media Server node only if you receive an Avaya factory installed WebSphereVoice Server machine, and you want to change the host name of that machine.

Prerequisites to IBM WebSphere Voice Server host name change procedures

• Install the WebSphere Application Server.

• Be familiar with the information in Change the Media Server Node host name on page 36.

• Be familiar with the appropriate IBM documentation.

• Obtain the IBM WVS DVD from Avaya.

IBM WebSphere Voice Server installation and uninstallation

68 Fundamentals — Speech MRCP 3 Dec 2010

Page 69: Fundamentals Speech MRCP

• Obtain a new host name from your administrator.

• Select a host name. The host name is the network name for the physical machine wherethe node resides. The host name must resolve to a physical network node on the server.Selecting a host name that other machines can reach in your network is extremelyimportant because the value that you specify for the host name is the value of the hostname property in WebSphere Application Server configuration documents.

Warning:Do not use the generic localhost identifier for this value.

IBM WebSphere Voice Server host name change proceduresThis task flow shows you the sequence of procedures you perform to change the host nameon the IBM WebSphere voice server. To link to a procedure, go to IBM WebSphere Voice Serverhost name change on page 68.

IBM WebSphere Voice Server host name change procedures

Fundamentals — Speech MRCP 3 Dec 2010 69

Page 70: Fundamentals Speech MRCP

Figure 2: IBM WebSphere Voice Server host name change procedures

IBM WebSphere Voice Server installation and uninstallation

70 Fundamentals — Speech MRCP 3 Dec 2010

Page 71: Fundamentals Speech MRCP

ResultIBM WebSphere Voice Server host name change procedures navigation

• Changing the host name in the WebSphere Application server on page 71

• Changing the host name on IBM HTTP server on page 72

• Changing the host name in WebSphere Voice Server with ifix IC51934 on page 72

• Updating the WebSphere Application Server Web Server plug-in on page 73

• Updating the Administrative Console Startup Link with the new host name on page 74

Changing the host name in the WebSphere Applicationserver

Change the host name on the WebSphere Application Server (WAS) by following thisprocedure.

1. Insert the Avaya WVS DVD into the DVD drive of the IBM WVS Media Server.

2. Copy the exportimport.zip file from the ifixes\misc directory on the DVDto the \bin directory on the IBM WVS Media Server where the WebSphereApplication Server resides.

3. Open a Command Prompt window by following the Windows path: Start, AllPrograms, Accessories, Command Prompt.

4. In the Command Prompt window, change the \bin directory as follows: cd c:\program files\WebSphere\AppServer\bin.

5. Unzip the exportimport.zip.

6. Check the directory cd c:\program files\WebSpher e\AppServer\bin\exportimport for the folder exportimport.

7. Open the exportimport folder by double-clicking on it.

8. Copy the file ValidateRelease.jacl from the exportimport folder to the\bin directory.

9. Copy the file exportImport.xml from the exportimport folder to the \bindirectory.

10. Stop the WebSphere Application Server following the Windows path: Start,Programs, IBM WebSphere, Application Server V6.1, Stop the Server menu.

11. In the Command Prompt window, enter the following command: ws_ant -buildfile exportImport.xml -logfile change Hostname.log -

Changing the host name in the WebSphere Application server

Fundamentals — Speech MRCP 3 Dec 2010 71

Page 72: Fundamentals Speech MRCP

DoldHostName=xxx -DnewHostName=yyy changeHostName where xxxrepresents the original host name and yyy represents the new host name.

12. Check the changeHostname.log file in the\bin directory for an error.

Changing the host name on IBM HTTP serverChange the host name on the IBM HTTP server by following this procedure.

1. Browse to the \conf directory where the IBM HTTP Server resides. The defaultlocation is: C:\program files\IBMHTTPServer.

2. Open the httpd.conf file using the Notepad application.

3. Edit the first line of the httpd.conf file, as shown in the following example. Replacetheoldhostname with the new host name you require:ServerName oldhostname # This is the main serverconfiguration file.

Changing the host name in WebSphere Voice Server with ifixIC51934

Change the host name of the IBM WebSphere Voice Server 6.1 (WVS 6.1) using theiFix_IC51934 by following this procedure.

This operation does not change the node name in the IBM Administrative Console, as hostname and node name are different fields and can have different values.

Prerequisites

You must obtain and read the contents of readme_IC51934.txt file for iFix IC51934 beforeyou change the host name in the WebSphere Voice Server with iFix IC51934.

1. Log on to the WebSphere Voice Server system as Administrator.

2. Create an ifix directory to contain the ifix_IC51934.zip file. For example,create the following directory to contain the iFix file:c:\download\WVSifixIC51934.

IBM WebSphere Voice Server installation and uninstallation

72 Fundamentals — Speech MRCP 3 Dec 2010

Page 73: Fundamentals Speech MRCP

3. Browse to the directory from step 2.

4. Unzip the ifix_IC51934.zip file.

5. Check to see if the WebSphere Application Server is started before proceeding. Ifthe server is not started, you start WebSphere Application Server by following theWindows path Start, Programs, IBM WebSphere, Application Server V6.1, Start theServer menu.

6. Begin the ifix_IC51934 installation script using the following command:ifix_IC51934.bat -installThe iFix install script provides status messages and directions.

7. Change to the \bin directory as follows:cd C:\Program Files\WebSphere\AppServer\bin.

8. Run the batch file changeWVSHostName.bat passing the new hostname asinput:changeWVSHostName.bat <newhostname>

9. Stop and restart the WebSphere Application Server.

10. Ensure that the machine is operational by running Verify Installation on the FirstSteps panel using the windows path Start, Programs, IBM WebSphere, VoiceServer v6.1.1, First Steps menu.

Updating the WebSphere Application Server Web Serverplug-in

Update the WebSphere Application Server plug-in by following this procedure.

1. Restart the machine using the Windows path: Start, Shut down.

2. Start the WebSphere Application Server using the Windows path: Start, Programs,IBM WebSphere, Application Server V6.1, Start the Server menu.

3. Enter the following URL address into the Internet Explorer address bar: localhost:9090/admin.

4. Click OK on the Administrator Login panel.

Updating the WebSphere Application Server Web Server plug-in

Fundamentals — Speech MRCP 3 Dec 2010 73

Page 74: Fundamentals Speech MRCP

5. Start the IBM HTTP Server using the Windows path: Start, Programs, IBM HTTPServer 1.3.28, Start HTTP Server menu.

6. Run Verify Installation on the First Steps panel using the windows path: Start,Programs, IBM WebSphere, Voice Server v6.1.1, First Steps menu.

Updating the Administrative Console Startup Link with thenew host name

Update the Administrative Console of the IBM WebSphere Voice Server 6.1 with a new nameby following this procedure.

1. Browse to the Administrative Console using the following path: Start , Programs,IBM Websphere, Application Server v6.1, Admin Console and right-click AdminConsole.

2. Select Properties.

3. Locate the Target section.

4. Change the original host name to the new host name.

5. Click OK.

6. Run Verify Installation on the First Steps panel using the windows path : Start,Programs, IBM WebSphere, Voice Server v6.1.1, First Steps menu.

IBM WebSphere Voice Server installation and uninstallation

74 Fundamentals — Speech MRCP 3 Dec 2010

Page 75: Fundamentals Speech MRCP

Chapter 8: Nuance Media Server installationand uninstallation

The following information provides instructions to install and to uninstall the Nuance Media Server in theMedia Applications Server 7.0 (Avaya MS 7.0) environment.

Navigation

• Nuance Media Server installation on page 75.• Uninstalling the Nuance media server on page 84.

Nuance Media Server installationThe following information provides instructions to install the Nuance Media Server in the MediaApplications Server 7.0 (Avaya MS 7.0) environment.

Prerequisites

• Obtain all required licenses from Avaya.• Back up current files and safely store them.• Satisfy disk space requirements. The amount of disk space Avaya software depends on

the node configuration, and the resource vendor specifies the disk space required for thevendor software of each Speech Server.

• Terminate all applications and close all programs on the nodes where you install software.• Disable any Web services such as Internet Information Service (IIS).

Installing the Nuance Media ServerThe following information provides instructions to install the Nuance Media Server in the MediaApplications Server 7.0 (Avaya MS 7.0) environment.

Fundamentals — Speech MRCP 3 Dec 2010 75

Page 76: Fundamentals Speech MRCP

Nuance Media Server installationThis task flow shows you the sequence of procedures you perform to install the Nuance MediaServer. You must perform each procedure in the order shown; otherwise, the installation canfail. To link to a procedure, go to Nuance Media Server installation navigation.

Nuance Media Server installation and uninstallation

76 Fundamentals — Speech MRCP 3 Dec 2010

Page 77: Fundamentals Speech MRCP

Figure 3: Installing the Nuance media server

Nuance Media Server installation

Fundamentals — Speech MRCP 3 Dec 2010 77

Page 78: Fundamentals Speech MRCP

ResultNavigation

• Installing the Nuance Recognizer 9 software on page 78

• Installing the Nuance Recognizer Language model on page 79

• Installing RealSpeak RS 4.5 on page 79

• Installing the RS 4.5 voices on page 80

• Installing the Nuance License manager on page 81

• Installing Nuance Speech Server 5.x software on page 81

• Installing the license file on page 83

• Restarting the Nuance Speech Server on page 84

Installing the Nuance Recognizer 9 softwareInstall the Nuance Recognizer 9.0 Recognizer software in the Media Applications Server 7.0(Avaya MS 7.0) environment by following this procedure.

Important:Install the U.S. English acoustic model with Nuance Recognizer resource software. To installanother acoustic model, you must install the Nuance acoustic model software for thatlanguage.

1. Browse to the<DVDDrive>: Nuance9.0 software directory.

2. Double-click Setup.exe to begin the installation.

3. Click Next in response to the Welcome window prompt.

4. Click Yes, I accept the terms in the license agreement button to accept the SoftwareLicense Agreement.

5. At the Destination Folder window, accept the default location or browse to theappropriate directory.

6. Click Next.

7. In the Ready to Install Program window, click Install.

8. Wait until the Setup window displays the status of the installation. Click Finish at theInstall Shield Wizard Completed window.

9. At the Restart Now window, select Yes.

Nuance Media Server installation and uninstallation

78 Fundamentals — Speech MRCP 3 Dec 2010

Page 79: Fundamentals Speech MRCP

Installing the Nuance Recognizer Language modelInstall the Nuance Recognizer Language model in the Media Applications Server 7.0 (AvayaMS 7.0) environment by following this procedure.

1. Browse to the <DVDDRIVE>: Nuance9.0, Languages directory.

2. Double-click the directory of the language you want to install.For example, if you want to install US English, double-click the English en-USdirectory.

3. Double-click the .msi file of the language model you want to install to begin theinstallation.For example, to install U.S. English, double-click Nuance Recognizer 9.0 - Englishen-US Language Pack.msi.

4. Click Next in response to the Welcome window prompt.

5. Click Yes, I accept the terms in the licensing agreement button to accept theSoftware Licensing Agreement.

6. At the Ready to Install the Program window, click Install.

7. Click Finish at the Install Shield Wizard Completed window. You do not have torestart.

Installing RealSpeak RS 4.5Install RealSpeak RS 4.5 in the Media Applications Server 7.0 (Avaya MS 7.0) environmentby following this procedure.

Prerequisites to installing RealSpeak RS 4.5Before you install RealSpeak RS 4.5 you must have at least 26 megabytes (26MB) of freespace on the server.

Installing the Nuance Recognizer Language model

Fundamentals — Speech MRCP 3 Dec 2010 79

Page 80: Fundamentals Speech MRCP

1. Browse to the <DVDDrive>: Realspeak4.5, RealSpeak_4.5_Windows softwaredirectory.

2. Double-click Setup.exe to begin the installation.

3. Click Next in response to the Welcome window prompt.

4. Click Yes, I accept the terms in the licensing agreement button to accept theSoftware Licensing Agreement.

5. At the Choose Destination Location window, accept the default location or navigateto the appropriate directory.

6. At the RealSpeak host service window, unselect Install the RealSpeak Host service.

7. Click Next.

8. At the Ready to Install Program window, click Install.

9. Click Finish at the Install Shield Wizard Completed window.

10. At the Restart Now window, select Yes.

Installing the RS 4.5 voicesInstall the RS 4.5 voices on the Nuance Media Server by following this procedure.

1. Browse to the <DVDDrive>: Realspeak4.5, voices directory.

2. Double-click the directory containing the voice you want to install.For example, if you want to install a male, U.S. English speaking voice, double-click the RealSpeak Host - American English - Tom directory. If you want to installa female, U.S. English speaking voice, double-click the RealSpeak Host - AmericanEnglish - Samantha directory.

3. Double-click Setup.exe to begin the installation.

4. In the RealSpeak Language Welcome window, click Next.

5. In the Ready to Install Program window, click Install.

6. In the Install Shield Wizard Completed window, click Finish. You do not have torestart.

Nuance Media Server installation and uninstallation

80 Fundamentals — Speech MRCP 3 Dec 2010

Page 81: Fundamentals Speech MRCP

Installing the Nuance License managerInstall the Nuance License manager on the Nuance Media Server by following this procedure.

1. Browse to the <DVDDrive>: NlicMgr11 directory.

2. Double-click Setup.exe to begin the installation.

3. The Nuance License Manager Welcome window appears. Click Next.

4. At the Destination Folder window, accept the default destination or click Change toselect a different destination folder.

5. Click Next to continue.

6. Click Install at the Ready to Install window.

7. Wait until the Installing License Manager window displays a progress bar indicatingthe progress of the installation.

8. At the Install Shield Wizard Completed window, click Finish.

9. Open the Nuance License Tools to configure the Nuance Licensing Service with theNuance license. To open the license tools, follow the menu path: Programs, NuanceLicense Manager, Licensing Tools.

10. Select the tab Service/License file.

11. Select Configuration Using Services.

12. Select Nuance Licensing Service.

13. Close the window.

14. Restart the server.

Installing Nuance Speech Server 5.x softwareInstall the Nuance Speech Server 5.x software on the Nuance Media Server by following thisprocedure.

1. Browse to the <DVDDrive>:NSS5.0 directory.

2. Double-click Setup.exe to begin the installation.

3. Click on Install Products.

Installing the Nuance License manager

Fundamentals — Speech MRCP 3 Dec 2010 81

Page 82: Fundamentals Speech MRCP

4. Select Nuance Speech Server.

5. Select the Custom Install window and install Apache.

6. Click Next in response to the Welcome window prompt.

7. Click Yes, I accept the terms in the licensing agreement button to accept theSoftware Licensing Agreement.

8. Click Next.

9. Confirm the user name and the organization.

10. Click Next.

11. Select Anyone that uses this computer.

12. Ensure that the following items are selected for installation.

• MRCP Server files

• RealSpeak Support

• Nuance Recognizer Support

• Nuance Verifier Server

• Apache

13. At the Destination Folder window, accept the default destination or click Change toselect a different destination folder.

14. Select Enable default language and select from the list of available languages.

15. Click Next.

16. Select the port and host name for Web sharing.The default values are: Web Host: localhost and Port: 80.

Important:For co-residency configuration, the port number for Web sharing must bechanged to the same port number as server.session.ossweb.port in theNSSserver.cfg file. For additional information about this parameter, see Configure the NSSserver.cfg file for co-residency on page 47.

17. Click Install.

18. Edit the %SWISRSDK%\config\Baseline.xml file and disable the parameterswirec_extra_nbest_keys . For additional information about this parameter,see Configure the Baseline.xml file on page 48.

Warning:You must disable this parameter before restarting the Nuance Speech Servernode.

Nuance Media Server installation and uninstallation

82 Fundamentals — Speech MRCP 3 Dec 2010

Page 83: Fundamentals Speech MRCP

19. Edit the %SWISRSDK%\config\Baseline.xml file to enable or disable theparameter swirec_extra_nbest_keys . For additional information about thisparameter, see Configure the Baseline.xml file on page 48.

20. Edit the %NSSSVRSDK%\config\NSSserver.cfg file and perform anynecessary site-specific or co-residency configuration. For additional informationabout the NSSserver.cfg file, see Configure the NSSserver.cfg file on page 45and Configure the NSSserver.cfg file for co-residency on page 47.

21. At the Restart Now window, clickYes.

Installing the license fileInstall a license file on the Nuance Media Server by following this procedure.

Prerequisites to installing the license fileStop the Nuance License Service from Windows services. If you do not stop the server theinstall fails.

1. Save the Nuance Recognizer 9.0.x license with the name nuance.lic in thefollowing folder: C:\Program Files\Nuance\license_manager\licence\.The full path name of the license file is: C:\Program Files\Nuance\license_manager\licence\nuance.lic.

2. Open Nuance Licensing tools by following the path: Start, All Programs, NuanceLicense Manager, Licensing Tools.

3. Select the Services/License file tab.

4. Select Configuration Using Services.

5. Select NuanceLicenseService.

6. Select the ConfigServices tab.

7. Examine the field path to the license file. This field shows the directory path wherethe Licenses Manager expects the license file to reside. It should display the defaultpath C:\Program Files\Nuance\license_manager\licence\temp.lic.

8. Edit the License File Path by changing \temp.lic to nuance.lic.

Installing the license file

Fundamentals — Speech MRCP 3 Dec 2010 83

Page 84: Fundamentals Speech MRCP

9. Examine the License File Path. The path to the license file should be C:\Program Files\Nuance\license_manager\licence\nuance.lic.

10. Select Save Services.

11. When the system prompts you to confirm the save, select Yes.

12. Close the window.

13. Start the Nuance License tool by following the path: Start, All Programs, NuanceLicense Manager, Licensing Tools.

14. Select the Start/Stop/Reread tab.

15. Click Start Server to start the server.

16. Use the Nuance License tool to perform a diagnostics check and verify that you cancheck out the license. The Diagnostics Check is found under the Server Diags tab ofthe License Tool.

Restarting the Nuance Speech ServerRestart the Nuance Speech Server by following this procedure.

1. To restart the Speech Server node, follow the path: Start > Restart the computer.

2. Navigate to Start > Settings > Control Panel > Administrative Tools > Servicesand ensure that the status of the Nuance Speech Server 5.0 service is Started .

Uninstalling the Nuance media serverThe following information provides instructions to uninstall software from the Nuance MediaServer . Use this information when uninstalling Nuance Recognizer 9, Nuance RecognizerLanguage model, RealSpeak 4.5, RS 4.5 Voices and Nuance Speech Server 5.x software.

Prerequisites

• Obtain a copy of the vendor documentation that relates to removing the Nuance MediaServer.

• Stop the services related to the speech product before you uninstall software and changethese services from Automatic Startup to Manual Startup.

Nuance Media Server installation and uninstallation

84 Fundamentals — Speech MRCP 3 Dec 2010

Page 85: Fundamentals Speech MRCP

1. Browse to Add or Remove Programs by following the path Start, Control Panel, Addor Remove Programs and double-clicking on Add or Remove Programs.

2. From the list in the Add or Remove Programs pane, select the vendor software thatyou wish to remove from the server.

3. Click Remove.

4. Click Yes to confirm the uninstall.

5. Follow the prompts to complete the process.

6. After the uninstall is complete, restart the server by selecting Start, Restart theComputer.

7. Deletes the Nuance folder from C:\Program Files manually.

Uninstalling the Nuance media server

Fundamentals — Speech MRCP 3 Dec 2010 85

Page 86: Fundamentals Speech MRCP

Nuance Media Server installation and uninstallation

86 Fundamentals — Speech MRCP 3 Dec 2010

Page 87: Fundamentals Speech MRCP

Chapter 9: Nuance Vocalizer installation

The following information provides instructions to install Nuance Vocalizer in the Media ApplicationsServer 7.0 (Avaya MS 7.0) environment.

Important:For a co-resident configuration, the order of installations is important. Perform the installations in theorder listed in the Navigation list below.

Navigation

• Installing Nuance Vocalizer software on page 87• Installing the Nuance Vocalizer voices on page 88• Installing the Nuance License Manager on page 88• Installing Nuance Speech Server software on page 89• Configuring Nuance Speech Server software on page 90• Testing Nuance Speech Server installation on page 92

Nuance Vocalizer installationThe following information provides instructions to install Nuance Vocalizer in the MediaApplications Server 7.0 (Avaya MS 7.0) environment.

Prerequisites

• Obtain all required licenses from Avaya.• Back up current files and safely store them.• Satisfy disk space requirements. The amount of disk space Avaya software depends on

the node configuration, and the resource vendor specifies the disk space required for thevendor software of each Speech Server.

• Terminate all applications and close all programs on the nodes where you install software.

Installing Nuance Vocalizer softwareInstall Nuance Vocalizer software in the Media Applications Server 7.0 (Avaya MS 7.0)environment by following this procedure.

Fundamentals — Speech MRCP 3 Dec 2010 87

Page 88: Fundamentals Speech MRCP

Important:You must have administrator privileges to install Nuance Vocalizer and the voice packs.

1. Extract the zip file containing Nuance Vocalizer to any folder.

2. Open Windows Explorer and navigate to the folder where the zip file was extracted.

3. Double-click Nuance Vocalizer For Network.msi to begin the installation.

4. Follow the on-screen instructions to install Nuance Vocalizer.

Installing the Nuance Vocalizer voicesInstall the Nuance Vocalizer voices by following this procedure.

Important:Repeat the following procedure for every voice that needs to be installed.

1. Extract the zip file containing the Nuance Vocalizer voice to any folder.

2. Open Windows Explorer and navigate to the folder where the zip file was extracted.

3. Double-click Nuance Vocalizer For Network en-US name.msi to begin theinstallation. Where name is the name of the voice to install.

4. Follow the on-screen instructions to install the Nuance Vocalizer voice pack.

Installing the Nuance License ManagerInstall the Nuance License Manager by following this procedure.

1. Extract the zip file containing the Nuance License Manager to any folder.

2. Open Windows Explorer and navigate to the folder where the zip file was extracted.

3. Double-click Nuance License Manager.msi .

4. Follow the on-screen instructions to install the Nuance License Manager.

Nuance Vocalizer installation

88 Fundamentals — Speech MRCP 3 Dec 2010

Page 89: Fundamentals Speech MRCP

5. Open the Nuance License Tools to configure the Nuance Licensing Service with theNuance license. To open the license tools, follow the menu path: Start, Programs,Nuance License Manager, Licensing Tools.

6. Select the tab Service/License File.

7. Select Configuration Using Services.

8. Select Nuance Licensing Service.

9. Select the tab Config Services Next to the field Path to the license file click onthe Browse button.

10. Select the license file received from Avaya.

11. Select the tab Start/Stop/Reread.

12. Select the buttons on the screen in the following order:

• Stop Server

• Start Server

• ReRead License File

Installing Nuance Speech Server softwareInstall the Nuance Speech Server software by following this procedure.

Important:The Nuance Speech Server software can be installed on both Windows and Linux platforms.The following procedure only describes the Windows installation.

Prerequisites

Before installing the Nuance Speech Server software, the following items are required:

• Windows 2003 server standard or enterprise edition. Only the 32-bit versions aresupported.

• Administrator privileges to install and configure the Nuance Speech Server service.• One of the following items (before installation of the Nuance Speech Server)

- Nuance Recognizer 9.0 (and a language pack)- Vocalizer 5.0 for Network (and a voice pack)

• Adobe Acrobat Reader 5.0 or higher (to read documentation)• Internet Explorer 5.0 or higher (to resolve URL MIME types)

Installing Nuance Speech Server software

Fundamentals — Speech MRCP 3 Dec 2010 89

Page 90: Fundamentals Speech MRCP

1. Extract the zip file containing the Nuance Speech Server to any folder.

2. Open Windows Explorer and navigate to the folder where the zip file was extracted.

3. Double-click NSS-5.1.0-i386-win32.exe to begin the installation.

4. Follow the on-screen instructions to install the Nuance Speech Server.The installer places the Apache 2.2.9 web server onto the system if it is not alreadyinstalled. At the same time, it modifies the Apache configuration file to include theSpeech Server web sharing configuration.

Important:If the installer detects Apache web server version 1.3 on the machine, it changesthe start type of the Apache 1.3 service from Automatic to Manual and installsApache 2.2.9. Version 2.2.9 is set up for Automatic start, but version 1.3 is stillavailable for any legacy applications.

5. After installation finishes, at the Restart Now window, click Yes.

6. Start the Nuance Speech Server service by either:

• navigating to Start, Administrative Tools, Services, selecting the NuanceSpeech Service service, and clicking on Start the service

• or by executing the command, startserver.bat from D:\ProgramFiles\Nuance\Speech Server\server\bin .

7. Open the log.txt file from D:\Program Files\Nuance\Speech Server\server\logs and make sure the server has started correctly.

Configuring Nuance Speech Server softwareConfigure the Nuance Speech Server software by following this procedure.

Important:The configuration of SIP signaling and uploading of custom applications is not discussed inthis document. Those procedures should be performed before the speech serverconfiguration.

1. In EM navigate to System Configuration, Signaling Protocols, MRCP, Serversand add a new MRCP server.

2. Using the table below, enter information in the fields on the Add MRCP Server pagein EM to configure the speech server.

Nuance Vocalizer installation

90 Fundamentals — Speech MRCP 3 Dec 2010

Page 91: Fundamentals Speech MRCP

3. Navigate to System Configuration, Signaling Protocols, MRCP, Servers, EditMRCP Server and add an MRCP resource.

4. Using the table below, enter information in the fields on the Add MRCP Resourcepage in EM to configure the speech server.

Important:Do not forget to add languages.

5. Navigate to System Configuration, Signaling Protocols, MRCP, Pools andcreate the pool.

Variable Definitions

Add MRCP Server

Field Name Field Entry

Server Name NSS

Server description Nuance Speech Server

Server address 192.168.109.47

Port 4900

Important:Be sure the Port number entered herehas the same value as MRCP transporton the Nuance Speech Server.

Vendor Nuance

Codec PCMU

State Unlocked

Add to Default Pool checkbox Checked

Add MRCP Resources

Field Name Field Entry

Server Name NSS

Vendor Name Nuance

Type menu TTS

URL Suffix media/speechsynthesizer

Weight 100

Configuring Nuance Speech Server software

Fundamentals — Speech MRCP 3 Dec 2010 91

Page 92: Fundamentals Speech MRCP

Maximum Sessions 100

Languages menu Add required languages

Testing Nuance Speech Server installationTest the Nuance Speech Server installation by following this procedure.

Using the example below, validate a call to the simple VXML application which usesthe Nuance Speech Server resources to play sound.

Example of the simple VXML application<?xml version="1.0" encoding="UTF-8"?> <vxml version="2.0"xmlns="http://www.w3.org/2001/vxml" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.w3.org/2001/vxml http://www.w3.org/TR/voicexml20/vxml.xsd"> <form> <block><prompt>Hello world!</prompt> </block> </form> </vxml>

Nuance Vocalizer installation

92 Fundamentals — Speech MRCP 3 Dec 2010

Page 93: Fundamentals Speech MRCP

Index

L

Loquendo Speech Server ............................................7

M

MRCPv1 .......................................................................7

Fundamentals — Speech MRCP 3 Dec 2010 93

Page 94: Fundamentals Speech MRCP

94 Fundamentals — Speech MRCP 3 Dec 2010