Hi, I am creating a pocketsphinx decoder to decode audio. In my grammar and
dictionary, "Operator, what time is it" is a valid command.
I capture audio of me saying, clearly, "What time is it", and pass it to the
decoder. It usually gets it right -- the first time I call it!
Subsequent calls are unpredictable... eventually I created a simple test
routine that passes the same audio, over and over, into the same ps_decoder.
And lo and behold, it gets different results each time. I thought that maybe I
was passing too much data at once to ps_process_raw, but I tried breaking it
into small chucks (same result). I thought maybe I'm missing some kind of call
to "ps_reset_decoder" or something, but I could not find any such API method.
Any idea why pocketsphinx would return different results for subsequent runs
on the same data? Here's my (very simple) C test function, and its results:
Hi, I am creating a pocketsphinx decoder to decode audio. In my grammar and
dictionary, "Operator, what time is it" is a valid command.
I capture audio of me saying, clearly, "What time is it", and pass it to the
decoder. It usually gets it right -- the first time I call it!
Subsequent calls are unpredictable... eventually I created a simple test
routine that passes the same audio, over and over, into the same ps_decoder.
And lo and behold, it gets different results each time. I thought that maybe I
was passing too much data at once to ps_process_raw, but I tried breaking it
into small chucks (same result). I thought maybe I'm missing some kind of call
to "ps_reset_decoder" or something, but I could not find any such API method.
Any idea why pocketsphinx would return different results for subsequent runs
on the same data? Here's my (very simple) C test function, and its results:
etc., etc.,
Hello
I can't reproduce your issue. Please provide a test case.