Speech application developers must often create grammars for common recognition tasks such as recognizing digit strings, telephone numbers, and currency amounts. The coverage of these grammars can have a significant effect on perceived system accuracy. As users interact with an increasing number of speech applications on a more frequent basis they will come to expect consistent system behavior even if these applications have been developed independently. Although a single, comprehensive definition of these tasks is unlikely to satisfy all requirements, a common vocabulary kernel may be within reach. In addition, many such grammars would benefit from translation and localization by residents of a particular target geography, or from tuning for specific speech recognition engines by acknowledged experts.
The Reference Grammar Repository Committee will work to establish the following:
- an initial collection of grammars for common recognition tasks
- a means for distributing these grammars
- a process for ongoing enhancement of existing grammars and contribution of additional grammars
- terms for incorporating these grammars in derivative works
- a user community to maintain and promote these activities
|End Date||31 August 2010|
2 Goals and Scope
The exploratory committee will define a framework for a standing committee that will establish and operate a reference grammar repository, including fundamental infrastructure, IP policy, and initial task scope.
The grammars should rely on relevant standards, such as SRGS and SISR. If deemed appropriate, this effort may draw on earlier projects towards creating common grammars, within both the VoiceXML Forum and other organizations.
Members should anticipate weekly conference calls lasting approximately one hour. Additional business may be conducted via the exploratory committee’s electronic mailing lists.
This charter was approved under the 30 May 2003 operating policies of the VoiceXML Forum. The terms of those operating policies apply.