V3 White Paper

VoiceXML Forum Technical Council

Rob Marchand, Editor

Overview

In the spring of 2004, the VoiceXML Forum conducted a survey of Forum members to collect feedback on the current and future direction of VoiceXML and related technologies.

One of the goals of this research was to provide the Voice Browser Working Group with additional industry direction regarding the evolution of VoiceXML, V3, and other standards in development by the VBWG.

Results

A summary of the results is as follows:

Thirty forum members responded;
Respondents included a mix of platform vendors, application developers, hosting providers, tools vendors, and training providers;
The majority of respondents are not involved with SALT at this time;
All are using or plan to use VoiceXML;
The value of VoiceXML as a standard is clear - as is the importance of its evolution.

A number of questions relating to VoiceXML 2.0, V3 and the evolution of VoiceXML were posed to the respondents. Results related to V3, are shown in Appendix A.

The VoiceXML Forum Technical Council believes that the key messages to the Voice Browser Working Group are as summarized below.

Most Important V3 Capabilities

Some level of compatibility with VoiceXML 2.0 is important - Respondents (25 out of 30) are highly interested in being able to have a clear migration path from VoiceXML 2.0 to V3 - whether by backwards compatibility, a VoiceXML 2.0 profile, or translation, this is viewed as useful.
Most respondents want multimodal markup to build on VoiceXML - Respondents (24 out of 30) are quite interested in leveraging their existing skills, VoiceXML capabilities, and the market acceptance of VoiceXML, in being able to develop a multimodal markup language based on VoiceXML, such as X+V.
The ability to communicate between VoiceXML and external entities is important - This capability has particular application in multimodal environments, and allows VoiceXML to communicate outside of the HTTP request/response model. Respondents (21 out of 30) indicated this ability was at least somewhat important.
Support for call control within VoiceXML itself is important - CCXML is viewed as an important standard - however, 20 out of 30 respondents indicated that call control capability within VoiceXML itself remains important.
Additional control over low-level media is desirable -Respondents (17 out of 30) wished to have a higher level of control over low-level media resources. These are available as extensions to VoiceXML 2.0 now in some cases, and a standardized mechanism for this is desirable.

Important V3 Capabilities

V3 should include support for Speaker Verification - Speaker Verification is viewed as a key requirement for V3 by 9 out of 30 respondents.
Although respondents are interested in modularization of VoiceXML, it is not yet clear which profiles should be defined - This is perhaps to be expected, as multimodal architectures continue to evolve (8 with no opinion, and others split among various profile options - client/server and voice/multimodal with 8 each).
Additional control over the FIA is desirable -Respondents (8 out of 30, with 10 N/A or don't know) indicated a desire to be able to exercise more control over the FIA, either via completely custom (pluggable) FIAs, or via a callback mechanism.

Summary

The VoiceXML Forum Technical Council has submitted these results to the VBWG as part of our mandate to provide industry-based technical input to the VoiceXML Forum, and to the W3C. It seems clear that VoiceXML is meeting many market needs today, but can continue to widen its appeal to industry. In addition, VoiceXML and related languages can act as valuable inputs to the V3 efforts.

For additional information regarding the VoiceXML Forum survey, contact the VoiceXML Forum Technical Council Liaison to the W3C, Ken Rehor (krehor@vocalocity.net).