So far, speech-enabling technology is starting to be utilized in contact center applications. VoiceXML (news -alert) and SALT (Speech Applications Language Tag) (news - alert) are both markup languages for implementing the speech interface as industry opens standards, allowing the developer to create a basic voice/speech application without having to know or learn anything about the voice hardware on which the application will run. They both have been creating wide opportunities in the marketplace. This article describes VoiceXML, SALT and how both works as well as how to implement a speech-enabled contact center application.

 

VoiceXML

 

VoiceXML tends to focus around the development of telephony-based applications and brings the advantages of Web-based development and content delivery to IVR (define - news - alert) applications. The specification allows developers to create audio dialogs that feature synthesized speech, digitized audio, recognition of spoken and DTMF (touch-tone) (news - alert) key input, recording of spoken input, telephony, and mixed initiative conversations. VoiceXML is designed to be platform-independent around the same server logic pull model used for HTML (define - news - alert) applications. In fact, VoiceXML applications can and often run in conjunction with traditional web applications, accessing the same data and performing the same essential tasks, even residing on the same machines.

 

SALT

 

SALT, Speech Application Language Tags, is actually a small set of XML elements and their associated attributes, events, and methods. These elements can be applied to existing markup languages such as HTML, XHTML, and WML to add speech and telephony interfaces to web applications.

The important concept in SALT is multimodal. As we have known about IVR, the majority of the application works using Speech/touch tone DTMF input and pre-recorded or synthesized speech TTS (define - news - alert) output. What we are really using here is a single modality "speech". Multimodality is where we can utilize more than one mode of the user interface with the application similar to our normal human communications with each other.

 

VoiceXML vs. SALT

 

Firstly, VoiceXML and SALT have different technical goals. Whereas VoiceXML is intended to focus around the development of telephony-based applications, it was developed to allow the specification of IVR applications in a markup language that leveraged the benefits of the World Wide Web. VoiceXML is a simple, high-level dialog markup language.

 

Secondly, SALT, a set of light-weight extensions to existing markup languages, tends to add speech-enabled telephony to web-based applications and turning them into multimodal model. SALT targets speech-enabled applications across whole devices including telephones, PDAs, (news - alert) tablet computers, and desktop PCs. However, VoiceXML focuses on telephony application development while SALT is focused on multimodal speech applications that can be accessed by a whole device. These points will help you to choose which one will be used in your contact center with speech-enabled technology.

 

The Weaknesses of VoiceXML and SALT In Building Contact Center

 

As the preceding description states, VoiceXML and SALT are both excellent standards for implementing speech-based applications. However, call control lacks in VoiceXML and SALT when using them to develop contact center applications, even if they have a couple of basic telephony functions, such as call transfer. As we have known in contact center applications, complex call control features would be used. There are two solutions: one is to utilize CCXML and another is to use CTI software.

 

CCXML Implements Call Control Features

 

CCXML, The Call Control eXtensible Markup Language, provides telephony call control that can be used in speech-enabled applications. CCXML can provide the call management, event processing, and conferencing capabilities that VoiceXML and SALT lacked. The following list represents some call controls features that CCXML can provide:

• Bridging: Connects a call between two call legs

• Routing: Routes inbound calls to the next available line in a group

• Conferencing: Multiple callers to join a voice conference

• Coaching: A third party to connect to a call, but only have one of the participants hear what is spoken

• Outbound Calling: Initiates a call and starts one or more dialogs once a connection is created

• Selective Call Answering: Decides whether or not to answer a call based upon caller�s information

Implementing Extra Call Controls Using Third Party�s CTI

 

If your contact center application requires complex call control functions, you can also implement them using third party�s CTI products by integrating speech platform with CTI software, such as Intel NetMerge CPS (formerly CT Connect), Genesys CTI and Cisco ICM CTI. You can customize many of the CTI features such as call routing, softphone, callback, screen pop, web chat, outbound, and conference.

 

How the VoiceXML and SALT-based Speech System Works

 

The speech system can be accessible by either voice caller or web user. The telephony interface containing telephony boards and interface software acts as a connector between voice caller and speech system. It is also responsible for detecting an incoming call, acquiring the initial VoiceXML/SALT document, and answering the call while the VoiceXML/SALT interpreter or voice browser conducts the dialog after answer. The web user visits speech system by web server that has deployed VoiceXML/SALT documents. Normally the web server processes requests from a client application and produces reply by VoiceXML/SALT documents over http. Also, the web server reponses the requests of interpreter or voice browser and speech engine which processes the TTS, prompts and grammars events. See Figure 1.

 

                                   

 

 

Microsoft Speech Server and IBM WebSphere Voice Server

 

There are a couple of VoiceXML and SALT based speech-enabling products in the marketplace.Obviously, the Microsoft Speech Server (MSS) 2004 and IBM WebSphere are leading providers in this field.

 

The Microsoft Speech Server (MSS) 2004 is based on SALT 1.0 and has been launched in March of this year. The MSS 2004 can be deployed as telephony (voice-only) and multimodal (voice/visual) applications that can be accessed by telephone, cell phone, desktop PCs, Pocket PC, Tablet PC and other devices. The tool MS Speech Application Software Development Kit (SASDK) is seamlessly integrated into Microsoft Visual Studio� .NET 2003 and provides a powerful set of ASP.NET speech controls, a Speech Add-in for Microsoft Internet Explorer, debugging tools, a speech application deployment service, tools for speech application log analysis, sample and reference applications, and a broad set of grammar library.

 

IBM WebSphere Voice Server 4.2 is VoixeXML-based speech platform. IBM WebSphere Voice Toolkit is a complete integrated VoiceXML application development environment. Voice Toolkit includes an integrated set of tools including application generating wizards, VoiceXML Editor, grammar development, CCXML Editor, testing and debugging tools, and a rich reusable dialog component. The IBM WebShpere Voice SDK and Toolkit are incorporated into WebSphere J2EE-based development environment.

 

The following table shows you a simple comparison between MS Speech Server 2004 and IBM WebSphere Voice Server V4.2.

 

 

Items	Microsoft Speech Server 2004	IBM WebSphere Voice Server V4.2
Markup Language	SALT 1.0	VoiceXML 2.0
Framework	MS .Net	J2EE
Programming	ASP.Net, C#.Net, VB.Net, Assembly	Java2, JSP, Servlet, CGI, JavaBean
Telephony Hardware	Intel DM/V480/960, D41JCT	Intel D/120/240/480/D300/D600JCT, Cisco/Siemens VoIP Gateway(H.323)
Speech Engine	Recognition/TTS/Prompt Engine Supported third party engine	Recognition/TTS/Prompt Engine Supported third party engine
SDK Toolkits	Yes	Yes
Telephony Simulator	Yes	Yes
Telephony Interface Software	Intel NetMerge Call Manager Intervoice TIM(Telephony Interface Manager)	C API provided for third party telephony platforms
Extra Call Controls	CSTA data extension	CCXML
Deployment in enterprise environment	Yes	Yes
OS Platform	Windows	Windows, AIX, Linux(Limited in features)
Speech recognition Language	US English, Beta for Multilanguage	US English/French/Germen/Spanish, etc. depending on platform
CTI(Computer Telephony Integration)	Not supported directly, but be able to integrate with third party	Not supported directly, but be able to integrate with third party

 

 

 

Implementing Speech-enabling Contact Center Application

 

 

Figure 2 is a typical architecture of VoiceXML, or SALT speech-enabling contact center. As we have seen, the speech server is able to integrate with PBX/ACD, CTI, CallRouting application (such as Genesys and Cisco ICM, etc.), CRM (such as Siebel, PeopleSoft, MS CRM, etc), desktop application, and data base. By using speech development tools to make available a speech application and perform unit testing in development server, you can deploy your speech application exactly as a web application, either J2EE or ASP.Net, in the web server.

 

The customers can access your speech application from a phone, cell phone, PDA and web. When a voice call comes in, the call is answered by speech server. At same time, the caller�s information, such as ANI, DNIS, CED, and channel of telephony boards will be retrieved by the CTI server and speech server. The customer interacts with the speech application by either spoken or DTMF by dynamic VoiceXML or SALT documents on the web server. Ultimately, the customer may need to talk to a live agent. The current call will then be transferred to an appropriate agent extension. According to the customer�s information retrieved and in conjunction with the business logics set, the CallRouting application runs a suitable routing script and instructs CTI and/or speech server to route the current call to an ACD queue. As soon as the extension of agent is available, the current call rings on the agent desktop. The CRM application pops up the customer�s messages and interacts with the database server.

 

Conclusion

 

According to your real requirements of contact center, you should choose VoiceXML or SALT and development an environment. You can also integrate a speech server with CTI, CallRouting application, and CRM into your real speech-enabling contact center application.

 

Xiaole Song is a professional on designing, integrating and consulting CTI, Contact Center, IVR, IP Telephony, CRM and Speech applications. He has performed various roles for Dialogic, Intel and Minacs Inc., etc. Feel free to email any comments about article or consulting services at [email protected]
 

 

Purchase reprints of this article by calling (800) 290-5460 or buy them directly online at www.reprintbuyer.com.

Respond to this article in our forums!