Update regarding speech model training inputs script? #13
Description
The documentation states:
Unfortunately, a key component - the script that generates the speech model training inputs and supporting data files - is currently in an unpublishable state. Nonetheless, with this excercise left to the reader, the align tool's help output explains its full usage. You may need to override CMUSPHINX_ROOT in the Makefile. Note that WAV files must be generated by FFMPEG because I hard-coded an offset to the audio data to avoid writing a RIFF parser.
Is there any progress on the script? It would be great to be able to run this locally. (The goal is to build a "Qurʾān Forge" of sorts - see https://github.com/quran/quran.com-api/issues/685. quran-align would serve as the automatic word timings file generator.)