[uiml] My report on Voice activities of W3C for today's OASIS UIML TC

[abrams <abrams@vt.edu>]

Attached is an informal summary I wrote for today's meeting on Voice 
activities at W3C...

-Marc

Marc Abrams
Co-Chair, OASIS User Interface Markup Language Technical Committee 
http://www.oasis-open.org/committees/uiml/

Adjunct Professor, Virginia Tech
Blacksburg, Virginia, USA
abrams@vt.edu

Co-founder, Harmonia Inc.
abrams@harmonia.com

540-357-0700
Information on Voice work at W3C
March 17, 2003
Marc Abrams

Reference:  www.w3.org/Voice/

GROUP #1:  Voice Browser Working Group

"The W3C Voice Browser Working Group (Members only) is defining a suite of markup languages covering dialog, speech synthesis, speech recognition, call control and other aspects of interactive voice response applications."  The following suite of languages forms the "W3C Speech Interface Framework":

- VoiceXML 2.0:  Describes aduio dialogs involving synthesized speech, digitized audio, recognition of spoken voice input, recognition of DTMF [the touch tone keys you press on a phone], and recorded spoken input.

- Speech Synthesis Markup Language (SSML): Markup language to describe voice prompts.  Allows specification of name, gender, age, pitch, speed, volume, and "emphasis".

- Speech Recognition Grammar (SRGS):  Describes a grammar (rules for parsing inputs) for DTMF and voice input.  Lets a grammar processor find a sequence of words that best matches what a speech recognizer hears when someone speaks.

- Call Control Markup Language (CCML):  Developers can use this markup langauge to screen calls, do call transfer, conditionally answer calls, place outbound calls, and similar functions.

- Semantic Interpretation:  I may have this wrong, but I believe SRGS describes SYNTAX, while "Semantic Interpretation" is a set of tags that specify semantic processing done when a grammar processor is trying to parse someone's voice input using SRGS.   (If you are familiar with compiler design for programming languages, then a parser uses both syntax and semantic rules.)

The Voice group wants to make these langauges work with XHTML, XForms, and SMIL.  There is a desire to combine speech p has been some controversy in W3C as to whether any member of W3C would have intellectual property rights on work created by the Voice group.  I believe that the result is the Voice group has been designated as a "royalty free" working group under a new patent policy at W3C.  (Incidentally, our TC may want to indicate on the TC's home page that the work of our TC is to lead to royalty free implementations of UIML.)

GROUP #2: Multimodal Interaction Activity:  I will investigate this for the April meeting.

======


Marc's comment on relation of Voice group to UIML:  
The vision of UIML is to simplify the world for user interface authors by permitting them to use one markup language (UIML), irrespective of target device.  The work of the Voice Browser Working Group to create multiple markup languages for voice.  The effort of the Voice group is very important to think through the many issues of how to build voice interfaces.  I feel that UIML plays a complementary role, becuase UIML can provide a way for user interface authors to use the Voice langauges that W3C develops without learning the syntactic details of those languages.  


The role of UIML then would be to provide vocabularies to support these voice markup languages.  In addition, organizations implementing UIML tools would then provide compilers or interpreters to implement a UIML document using these voice vocabularies using the Voice Browser Working Group's new languages.

If we step back a moment and look at the overall work at W3C to develop many markup languages to carefully describe user interfaces for voice, web browser, and other platforms, then the need for UIML becomes even stronger.  User interface authors are faced with an increasing number of implementation languages (such as the ones from the Voice browser group), and need the simplicity of one language.