Create Test and Train folders in the Audio folder and store the audio data.
Create Test and Train folders in the Data folder with the following files in each folder.
- spk2gender
<speakerID><gender> - wav.scp
<uterranceID> <full_path_to_audio_file> - text
<uterranceID> <text_transcription> - utt2spk
<uterranceID> <speakerID>
Create a folder local with the following file and a subfolder dict -
- corpus.txt
<text_transcription>
Create the follwing files in dict
- lexicon.txt - Can be obtained from http://www.speech.cs.cmu.edu/tools/lextool.html
<word> <phone 1> <phone 2> ... - nonsilence_phones.txt
<phone> - silence_phones.txt -
sil
spn
- optional_silence.txt
sil
Copy the utils and steps folders from Kaldi-Master/egs/wsj/s5 into the model directory
Copy score.sh from Kaldi-Master/egs/voxforge/s5/local into the local folder of the model directory.
Create a folder conf in the model directory and copy the decode.config and mfcc.conf files from this repository.
Copy the files cmd.sh, path.sh and run.sh to the project directory
Run ./run.sh