I'm playing with pocketsphinx with continuous recording and silence detection.
Occasionally it trips to E_FATAL in
pocketsphinx-0.8/src/libpocketsphinx/acmod.c
"Decoder can not process more than 32767 frames at once, requested 51200"
The requested number is typically around 50000.
I wonder if I'm missing a call to clear something occasionally, or what is wrong?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
By default pocketsphinx is configured to process only approximately 5min of audio per utterance. You need to make sure your voice activity detection doesn't produce longer utterances. In VAD code you need to end utterance before 5 mins expire.
Otherwise, you can reconfigure pocketsphinx to decode longer utterances. You can do it in hmm.h by fixing the following variables:
I'm playing with pocketsphinx with continuous recording and silence detection.
Occasionally it trips to E_FATAL in
pocketsphinx-0.8/src/libpocketsphinx/acmod.c
"Decoder can not process more than 32767 frames at once, requested 51200"
The requested number is typically around 50000.
I wonder if I'm missing a call to clear something occasionally, or what is wrong?
By default pocketsphinx is configured to process only approximately 5min of audio per utterance. You need to make sure your voice activity detection doesn't produce longer utterances. In VAD code you need to end utterance before 5 mins expire.
Otherwise, you can reconfigure pocketsphinx to decode longer utterances. You can do it in hmm.h by fixing the following variables:
to
Last edit: Nickolay V. Shmyrev 2013-04-05