Hello Alberto,
First, the changes I made in the demo were around UIO recording; the buffer size was initialized with unsupported value, and the recorder didn't know what to do with it. It worked fine with VS 6.0, though : -)
Now, as far as your app goes - I'm not sure I like the way the CSP parameters are set:
04/10/2016 23:09:06.635 5716 5948 dm3voice DEBG CspChannel dxxxB1C1 ----- NLP=1SVAD=0 SpchSnr=0 SpchThresh=-40 SpchTrig=10 SpchWindow=10 Dur=0 Len=128 SCSOnOff=0 SCSTrailSil=200 SCSInitData=0
Without SVAD the streamer may be waiting for bargein to start, but this is also disabled:
04/10/2016 23:09:34.574 5716 2476 dm3voice DEBG CspChannel dxxxB1C1 ----- VADEvt:Off Converged:Off BargeIn:Off
So, the streamer is ready, but doesn't know when and why to start
For HMP I'd recommend the following settings:
// Enable Barge In
param = 1;
ec_setparm(m_hDevHandle, DXCH_BARGEIN, (void *) ¶m);
param = 1;
ec_setparm(m_hDevHandle, DXCH_BARGEINONLY, (void *) ¶m);
param = 1; // enable VAD
ec_setparm(m_hDevHandle, ECCH_VADINITIATED, (void *) ¶m);
param = 1; // enable both energy and zero-crossing methods for VAD
ec_setparm(m_hDevHandle, ECCH_SVAD, (void *) ¶m) ;
param = 0;
param = -30; // -40 also works OK, but masy cause mistrigger on minor line noise.
ec_setparm(m_hDevHandle, DXCH_SPEECHPLAYTHRESH, (void *) ¶m);
Leave NLP in its default state; if you don't have problem with echo, NLP is not needed.
If this still does not work, please try to use pure recording instead of ec_stream; for this, of course, you'll need to comment out dx_play. Some sort of simple scenario - ipm_StartMedia / dx_reciottdata. To make sure you have the audio on input.
If this still