Forced alignment with Sphinx?

Speech Recognition Toolkit

Brought to you by: air, arthchan2003, awb, bhiksha, and 5 others

This project can now be found here.

Forced alignment with Sphinx?

Forum: Help

Creator: Jamie Bullock

Created: 2014-02-19

Updated: 2014-02-19

Jamie Bullock - 2014-02-19

Hi,

The problem I am trying to solve is this:

I have a collection of audio files containing dialog for an arbitrary number of speakers all in American English

The audio files are clean and only contain dialog, nothing else

For each audio file I have a text file that is a word-for-word transcript of the dialog

I need to label each word in the transcript with its temporal location in the audio file

Ideally I also need a temporal location for each phoneme within a word

My application is written in C, so I need to solve the problem with C/C++ libraries and / or custom code.

Is this something that Sphinx can be used for?

I have read online that "Sphinx 4" can be used for this, but it appears to be a Java library.

Is it possible to solve this problem using Sphinxbase with additional custom C code?

Sorry if these questions are obvious, I am new to the Sphinx ecosystem!

Thanks,

Jamie
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Nickolay V. Shmyrev - 2014-02-19

How long each file is?

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Log in to post a comment.