I don't know if this is possible to do with Sphinx, but it looks feasible (from a non expert point of view).
I have a set of speech audio files (clean, recorded in studio), and I also have available the transcription of those audios, in txt format.
What I'd like to do is use Sphinx to know what exact time (hh:mm:ss.cc) of every word in transcription. I guess that, as we only should check for a limited amount of words, we can get a high accuracy, but honestly, I have no remote idea on where should I start.
Can anyone point me in a good direction ?
Regards,
Albuendi
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
It depends on the amount of text you want to syncronize. If it's a short sentence, we have forced alignment mode. If it's dozen of minutes, we have the following tool:
I was looking for this months ago, but could find nothing. I see now that announcement date is July 27th, so I didn't se in my first investigations, sorry about that.
Great thing, as announcement says.. I'll use this for subtitle synchronization, so it's perfect for our use case !
Regards,
Last edit: Alfonso Buendía 2014-09-06
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello,
I don't know if this is possible to do with Sphinx, but it looks feasible (from a non expert point of view).
I have a set of speech audio files (clean, recorded in studio), and I also have available the transcription of those audios, in txt format.
What I'd like to do is use Sphinx to know what exact time (hh:mm:ss.cc) of every word in transcription. I guess that, as we only should check for a limited amount of words, we can get a high accuracy, but honestly, I have no remote idea on where should I start.
Can anyone point me in a good direction ?
Regards,
Albuendi
It depends on the amount of text you want to syncronize. If it's a short sentence, we have forced alignment mode. If it's dozen of minutes, we have the following tool:
http://cmusphinx.sourceforge.net/2014/07/long-audio-aligner-landed-in-trunk/
You need to checkout latest pocketsphinx to use it. Let me know if you have issues.
Thank you very much Nickolay !
I was looking for this months ago, but could find nothing. I see now that announcement date is July 27th, so I didn't se in my first investigations, sorry about that.
Great thing, as announcement says.. I'll use this for subtitle synchronization, so it's perfect for our use case !
Regards,
Last edit: Alfonso Buendía 2014-09-06
Ok, let me know how it works.