tags:

views:

28

answers:

2
Intially I have specified MCI_WAVE_SET_PARMS at the time of recording as follows:

MCI_WAVE_SET_PARMS mciSetParms;
mciSetParms.wFormatTag = WAVE_FORMAT_PCM;
mciSetParms.wBitsPerSample = 16;
mciSetParms.nChannels = 2;
mciSetParms.nSamplesPerSec = 11050;

Now if I change it to 

MCI_WAVE_SET_PARMS mciSetParms;
mciSetParms.wFormatTag = WAVE_FORMAT_PCM;
mciSetParms.wBitsPerSample = 8;
mciSetParms.nChannels = 1;
mciSetParms.nSamplesPerSec = 8000;

Will it affect the speech recognition performed by SAPI ?

A: 

I think it should affect .

See the following link :

http://en.wikipedia.org/wiki/Acoustic_Model

Ashish
+1  A: 

Yes, it will. SAPI's recognizer really needs at least 11KHz of 16-bit audio for good recognition. Giving it 8KHz and 8-bit audio will substantially impair the recognition rate (if you get any recognitions at all).

Note - Dropping the channels from 2 to 1 won't affect SAPI at all.

Eric Brown