0
audio to text conversion is very slow. why?
Want to covert audio .wav files to text using python v3 Sphinx module, its working fine but problem is it's taking long time for large .wav files and I have attached my code here. Please anyone let me know for any changes to work faster and it will be helpful Is there any alternative ways to convert audio to text in offline? https://code.sololearn.com/c42j8G8cE176/?ref=app
3 Antworten
+ 3
Do you know what line is choking in your code? (I see the block timer, I mean timing them individually to narrow down a suspicion?)
Also, I was just using PocketSphinx for speech recognition directly:
https://code.sololearn.com/c2f82jygtyK6/?ref=app
Note my parameter list; I think buffer size is important (but it's been a while; others may matter for speed too)
0
Kirk Schafer I executed your code by adding my audio file path but output is irrelevant and it is not matching up single word of audio file.
0
I used a carefully-selected file. You probably need to tune or train it (in the right ways):
Training an acoustic model...
https://cmusphinx.github.io/wiki/tutorialam/
Tuning speech recognition accuracy:
https://cmusphinx.github.io/wiki/tutorialtuning/
"...[determine if] accuracy is just lower than expected or whether it is very low in general. If the accuracy is very low in general, you most likely misconfigured the decoder. If it is lower than expected..." [create training database, match sample rate/channels], etc
The parameters I'm using come from (+ usage samples):
https://pypi.org/project/pocketsphinx/