Hello all,

I've been working on a small project which involves recording audio then recognizing it afterwards.  The system works perfectly if I use _continuous through the microphone, but fails when using -infile.  _batch also fails somewhat differently.  I have been using ffmpeg to convert the audio to pcm headerless 16k mono sound but the recognition is still completely wrong.

Here are the commands that I am running to make this work

ffmpeg -y -i ..\files\test.wav -f s16le -acodec pcm_s16le -ar 16000 -ac 1 test.raw

pocketsphinx_continuous  -argfile argfile.txt -infile test.raw

Can anyone point me in the correct direction?

Thanks,

Geoff

--- (Edited on 4/17/2013 9:22 am [GMT-0500] by v2geofs) ---