Dmitriy - 2014-11-04

Hello guys,

I faced to the issue and need to understand can Sphinx help me to solve
it or not.

There is already existing on my Test environment. It is includes
Asterisk 1.8.10 running on the Ubuntu Linux + uniMRCP (same machine) +
Nuance speech Server running on CentOS.

Current issue:

1) When it is call and MRCPSynth extension starts to work, the Nuance
produces TTS and for example says "Press 1 to subscribe for
out channel":
- if the user press 1 during the speech it will interrupt the
speech and the MRCPSynth extension will return 0 (in any case pressing
1, 2, 3 .... )
- so user need of to wait until MRCPSynth extension finish it's
work and then press the "1", or press it to times to be recognized, and
this confused many users.

2) I have found that it is needed to use SynthAndRecog extension in the
Asterisk to be able
recognize DTMF tone while the voice menu is spoken up.

3) However it is needed to have (as I understand) external MRCP ASR
recognizer

So I need a simple recognizer which just can recognize DTMF tones. Does
Sphinx can help me with this issue, or there is different solutions for
this.

I am new to this area, I am java dev currently, but need to made the
proof of concept,

Thanks in advance,
Dmitriy