Audio Search

[SKR]

Hi all,

Is audio search (spoken word) possible with pocketsphinx??

I found keyword search with pocketsphinx, given a keyword with a threshold search against an audio file.

for ex:
pocketsphinx_continuous -infile sample.wav -hmm ./ -dict msr4search.dic -lm msr4search.arpa -kws keyphrase.list -time yes -logfn result.log
results keyword along with timestams.

Here 'keyphrase.list' contains some text input. Can I place some audio files instead of text input in keyphrase.list??

Or anyother solutions to audio search??

Thanks in advance....

[Nickolay V. Shmyrev]

Transcribe speech and look for the required phrases in transcription.