WebSphere Voice Response for AIX

At a glance

WebSphere Voice Response for AIX, V4.2 features:

• Network Equipment Building Standards (NEBS) compliant

• BladeCenter® support

• Enhanced Telephony Protocol support for next-generation voice networks — Voice over Internet Telephony using Session Initiation Protocol (SIP)

• Service provider networks: SS7 (Signalling System #7) T1 (ANSI Signalling System #7) and E1 (ITU-ISUP Generic)

• Enterprise networks — QSig, R2MFC Korean

• Call Control XML (CCXML), V1.0 support

• Enhanced VoiceXML, V2.1 scalability to 480 channels/system

• Interoperability with WebSphere Voice Server, V4.2 and V5.1 (Voice XML only) — for conversational speech applications

• Interoperability with WebSphere Voice Application Access, V5.0 for Voice Portal applications.

• Improved echo cancellation for speech recognition-enabled applications

• In-site migration from WebSphere Voice Response for AIX, V3.1

• New fax card compatible with the latest IBM pSeries™ machines

• Fully accessible WebSphere Voice Response when running CCXML and VoiceXML, V2.1 applications

WebSphere Voice Response is a proven platform for delivering solutions to service providers:

• An open platform running UNIX with CCXML and VoiceXML industry-standard programming environments as well as IVR programming environments for its existing customers.

• Resilient and highly scalable, with its pSeries, PCI telephony adapters, and AIX industrial strength support meeting NEBS compliance.

• Highly scalable to 480 channels per pSeries system. Such systems can be clustered for large installs.

• Improved highly scalable and redundant SS7 software is also provided in this release when connected to Telco switches.

• Supports next-generation SIP networks with existing applications and services. Access to new signalling information allows development of new and enhanced services.

• Mass-calling applications are supported.

• Scalable, redundant, and resilient centralized application management using application servers.

• IBM HACMP and WebSphere Voice Response Single System Image (SSI) clusters for state table and custom server applications.

What's new in V4.2:

Support for Network Equipment Building Standards (NEBS) compliant BladeCenter

• WebSphere Voice Response for AIX Version 4.2 can be used on a NEBS compliant BladeCenter computer for Voice Over Internet (VoIP) Telephony capability using Session Initiation Protocol (SIP).

Additional hardware adapter cards are no longer required to run WebSphere Voice Response for AIX

• WebSphere Voice Response for AIX Version 4.2 can be used without the need for any adapter card on a BladeCenter or System p5™ computer for Voice Over Internet (VoIP) Telephony capability using Session Initiation Protocol (SIP).

AIX support upgraded to Version 5.3

• WebSphere Voice Response for AIX Version 4.2 supports the use of AIX Version 5.3 as its operating system. AIX Version 5.2 is the minimum level that can be used when using an adapter card.

Significant enhancements in VoIP telephony using SIP are incorporated in WebSphere Voice Response for AIX, V4.2. V4.2 uses the DTNA software adapter implementation to connect to a VoIP network.

VoIP features include:

• BladeCenter or System p5 computers can handle eight trunks and simulate up to 240 SIP endpoints

• DTNA software adapter implementation supports Real Time Protocol/Real Time Control Protocol (RTP/RTCP) over 100 Mbit/s Ethernet for User Datagram Protocol (UDP) packetized voice data

• DTNA can coexist with Digital Trunk Extended Adapter (DTXA) or DTTA in the same system unit allowing hybrid PSTN/VoIP configurations in, for example, IP Call Centers

• DTNA supports uncompressed G.711

• DTMF send/receive keys using RTP payload packets — RFC2833 (telephone event)

• SIP protocol signaling support using system Ethernet port

• SIP protocol support over TCP or UDP as per RFC 3261

• Inbound call, outbound call, and blind and supervised call transfer supported

• SIP connections to Soft phones, Hard phones, and proxies which conform to SIP RFC 3261

• Existing and new state tables, Java and VoiceXML, CCXML applications can be programmed to handle SIP calls

• Calls can be tromboned between DTNA channels using Trombone Custom Server. Tromboning between PSTN and SIP channels is not supported

• Additional interface between SIP and applications using "Tagged Strings" (Diversion Header, From Header, To Header, Request Header, Alang Header, Final Response, Subject Header)

• Reinvite from remote party allowing mid-call codec changes

• Subject Header on Bye supported for end-of-call application to gateway information transfer

• Blind notify for Message Waiting Indication (MWI) control

• MWI manual table-driven DNS SRV support for MWI allowing configuration of primary and secondary proxies, and proxy pooling

Service Provider Networks: SS7 features

• Reuse of E1 hardware adapters from previous solutions.

• Improved scalability from 1400 to 2304 (T1) or 2880 (E1) channels.

• Support for previous SS7 function with ITU-ISUP Generic as main switch connection.

• Support for T1 ANSI (generic and MCI variant) networks.

• Support for E1 ITU 1997, ITU White Book (1992, 1993), CCITT Blue Book (1988).

• Standard T1 or E1 link is for signaling; no support the for the older serial V35, with insufficient bandwidth for 2000 channels. Newer VoIP transports are not supported.

• The SS8 D7 product has a different redundancy strategy from the previous version. Two links, SS8 cards, and D7 stacks are both active (load-sharing) at all times so if one is lost, the other is ready with no loss of calls.

Enterprise-based signalling extensions: QSig: WebSphere Voice Response, V4.2 supports inbound and outbound calls and message waiting indicator using the QSig protocol.

CCXML, V1.0 support

• Use of CCXML is optional. Existing applications not using CCXML do not need to change.

• CCXML enables XML-based call control for Enterprise and service provider applications and enables all applications to be stored on central Web/application server for reliability and redundancy.

• CCXML enables the routing of incoming calls to specific applications based on ANI, DNIS simplifying the configuration on base Voice Response platform.

• CCXML can be used with VoiceXML, V2.1 and Java applications with basic inbound and outbound calls handled by CCXML.

• The CCXML Browser will parse all tags according to the level to W3C standards it supports.

• The WebSphere Voice Response signalling process can pass extra protocol data, such as ISDN and SS7 information elements and SIP Web sites (in tags using ECMA script) on incoming calls to CCXML applications.

• The WebSphere Voice Response signalling process can present calls in alerting to CCXML application where protocol permits, enabling playing of announcement messages to caller without the caller being connected and caller being charged (including early media support).

• CCXML enables specification of a channel group on an outbound call to select which protocol to make the outbound call, that is, PSTN like ISDN, SS7, or on VOIP like SIP.

• Use with CTI products for contact center, call center applications, and solutions.

• CCXML is responsible for call handling within an application with the ability to call out to CTI products.

The WebSphere Voice Toolkit provides a CCXML editor as the recommended method of creating CCXML applications. It enables CCXML to be developed using the same tooling as VoiceXML and Speech Grammar development.

Enhanced echo cancellation: Echo cancellation is on the DTTA adapter improving capability and reducing cost, without the need for external echo canceller boxes. Enhancements include the following:

• Continuous convergence and cancellation, instead of initial convergence required by the previous canceller

• Removes the need for an initial force play convergence prompt

• Improved echo cancellation during the call on account of the continuous convergence up to 32 msec echo delays

• Removes possibility of failure of initial calibration due to excessive echo (currently, echo cancellation within the call is unusable if the initial calibration fails)

• Improved overall echo cancellation (smaller echo residual signal means improved voice recognition and more reliable barge-in)

Enhanced scalability of VoiceXML to 480 channels with DTTA adapters

• Simple VoiceXML V2.0/2.1 applications, with no speech recognition or text to speech, can be supported on a single server with four 1.5 GHz processors.

• More complex applications with speech recognition or text to speech can be supported with 480 channels of telephony on a single server but with VoiceXML V2.0/2.1 browsers and WebSphere Voice Server systems distributed across other servers.

Additional on-demand feature: Customer-owned licenses can be managed using License Use Management (LUM) tools. Licenses can be shared across multiple servers and sites. This allows customers to be responsive to immediate demands on their systems and allocate channels on systems which need them. The licenses are enabled when the channels are enabled.

The Voice Toolkit is no longer included on CD in the WVR 4.2.3 Media Pack. You can download the latest version from the WebSphere Voice Toolkit Web site:

http://www.ibm.com/software/pervasive/voice_toolkit

The toolkit supports the CCXML 1.0 and the VoiceXML 2.1 specifications, and includes a grammar editor, pronunciation builder, and an audio recorder. You can configure a development environment to create, test, and debug custom voice portlets using VoiceXML 2.0 or 2.1. Other features of the toolkit include:

• The ability to debug your portlets using the local debugging environment

• The ability to create VoiceXML applications using the new Communication Flow Builder

• An editor that can handle both CCXML and VoiceXML source code

• A conversion wizard to assist you in migrating any VoiceXML 1.0 applications to 2.0 or 2.1

• An integrated VoiceXML 2.1 Application Simulator and Debugger

• Integrated concatenative text-to-speech (CTTS) and speech recognition engines

The toolkit editor also includes a wizard that allows you to select and customize Reusable Dialog Components (RDC) written to the VoiceXML 2.0 or 2.1 specification. These RDCs contain pretested code for commonly used functions such as credit card type, currency, date information, and so on.

The WebSphere Voice Toolkit V6.0 is enhanced to support the latest VoiceXML V2.1 specification.