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

[Kevin P. Fleming]

Josh McAllister wrote:
> I sense from the "short" tone of your response that you feel this to be
> off topic here. Is there a better forum for me to start a dialogue about
> this? I understand some of the questions could have been answered myself
> with enough time spent sifting through code, but does that make the
> entire post irrelevant? Is MRCP destined to be completely irrelevant to
> asterisk dev discussions? Honestly, I have no desire to start a flame
> war, especially not with someone as highly esteemed in the * community
> as yourself. But I am truly confused by this reply and the dead end road
> it sent this thread down.

There was no intention to turn you off, or to be 'short'. However, the
existing code is -extremely- straightforward, and is actually documented
adequately in speechrec.txt that it would have answered your initial
questions quite well. When we see someone post on this _developer_ list
and they haven't done their homework, it makes us assume the worst :-)

In no way is MRCP 'irrelevant', nor did I state that at all. If someone
wants to build an MRCP 'connector' for the existing speech recognition
API, they are more than welcome to do so. I have no idea how the MRCP
API works, so I can't speak to any potential licensing issues that may
arise, but that will have to be kept in mind as well.

There is no generic text-to-speech API at this point, and we have not
yet managed to release our 'native' Cepstral module either (in spite of
the fact that Cepstral is certainly one of our partners). The reason
that I said you should contact Digium Sales is because the development
that needs to occur to complete that project will be more likely to
happen if the sales department is pushing for it.