Since, we will have audio recordings segmented, and saved onto the server, we will spawn a new thread (or process) for each segment in order to perform pocketsphinx_continuous on the wav file. This will require us to modify the existing converter_wav_text script:
...
# apply CMUSphinx on audio wav file
inotifywait ../audio/recording_converted -m -e close_write |
while read path action file; do
# do something with the file
filename="${file%%.*}"
pocketsphinx_continuous -infile ../audio/recording_converted/"$file" -hmm en-us -lm cmusphinx-5.0-en-us.lm.dmp >| ../audio/recording_text/"$filename"
done
Since, we will have audio recordings segmented, and saved onto the server, we will spawn a new thread (or process) for each segment in order to perform
pocketsphinx_continuous
on the wav file. This will require us to modify the existingconverter_wav_text
script: