Respected sir,
I am getting a high accuracy when I use "-inmic" argument to record from
device microphone but when I use the same recording as a .wav file and use
"-infile" argument, I am getting a low acuracy.
Why is this happening ? Also, is there any specific format in which the
audio is fed as input to pockesphinx when I use "-inmic" ? Is this a
.raw/.pcm format unlike .wav format ? Looping the .wav file once to get cmn
values for the device works fine and when second time iteration comes the
output is much more accurate. Is there a way to deal with this looping of
.wav file ?
Respected sir,
I am getting a high accuracy when I use "-inmic" argument to record from
device microphone but when I use the same recording as a .wav file and use
"-infile" argument, I am getting a low acuracy.
Why is this happening ? Also, is there any specific format in which the
audio is fed as input to pockesphinx when I use "-inmic" ? Is this a
.raw/.pcm format unlike .wav format ? Looping the .wav file once to get cmn
values for the device works fine and when second time iteration comes the
output is much more accurate. Is there a way to deal with this looping of
.wav file ?
Regards,
Mainak Biswas