[Speechsc] VoiceXML Input-modes problem

Hi,
The topic of meeting VoiceXML input-modes property requirements through
MRCP has come up a few times before.

One part of the problem(relating to barge-in support) was addressed by
the following
    1. change START-OF-SPEECH to START-OF-INPUt
    2. Add support for Input-Type header into the START-OF-INPUT event,
which specifies what type of input was detected.

The second piece relating to controlling the recognition engine to limit
recognition to dtmf-only or speech-only was not addressed.

I would like to address this problem by proposing a new header
"Input-Mode" header. This will tell the recognition engine whether it
should recognize speech only, dtmf only or both. The possible values for
this header are "speech-only", "dtmf-only" and "speech-dtmf". This
header MUST be supported by a recognition engine and MAY be carried in
the RECOGNIZE, SET-PARAMS and GET-PARAMS methods and their responses. 

Note: This is a proposal for resolving this issue. It does not in anyway
mean that this header/solution will or will not get into the next draft.
That would depend purely on the time and resource constraints of the
editors in releasing the next draft and the fact that the document is
already in last call and the change limitations it might impose. If it
does not get into the next draft, which is likey, this and a few other
resolved but pending proposals will be published as a follow-on
draft/RFC.

Thanks,
Sarvi  

[Speechsc] VoiceXML Input-modes problem - A proposal