[asterisk-dev] Asterisk + MRCP for Speech Resources (TTS / ASR) ?

[Kevin P. Fleming]

Josh McAllister wrote:
> Furthermore there is some work going on over at
> http://www.speechforge.org/ to provide an OSS MRCP implementation called
> Cairo. Cairo is currently capable of wrapping FreeTTS and Sphinx4. I
> know a lot of work has already been done to integrate LumenVox, but I do
> wonder what direction was taken, and if there is any possibility of
> steering this way? 

'what direction was taken' is clearly evident from the code... We
created a generic Speech Recognition API, that understands only high
level concepts like 'grammars', 'audio' and 'results', and does not care
what the format of any of those are.

> Alternatively, would it be feasible to develop an * <-> MRCP plugin /
> middleware using the existing Speech API? It seems like that might
> bridge the gap for now, and really open the door for users to choose
> from a variety of ASR software.

Of course it would be possible.

> Side question -- I saw that Cepstral is now listed on Digium Partners
> page, have I missed the announcement for integrated support, or is this
> a hint of something to come? If this has not yet been implemented, has
> any thought been given toward implementing via MRCP?

That is off-topic for this list; questions about commercial products or
Digium Partners should be directed to the Digium Sales department :-)