Hello, I recently stumbled upon this tool as I'm trying to train Coqui on my own voice. However, I can't seem to get the program to work as intended, with the two main issues mentioned in the title.
Issue 1 - 'Immediately' Ignoring Longer Files (>5 min)
The program would split the files successfully but would output nothing in the zip, I had to modify silence threshold for it to work, as the original configuration would literally immediately ignore it
Issue 2 - Shorter files are not being split in the zip
I've tried uploading several smaller files and they seem to get split, transcribed. However, the zip output is just my original wav file and a single line of transcription, just for that file.
Edit: By 'successful split', I mean it get outputted to 'wavs_split_final' folder.
I have no idea what I'm doing wrong, I'm running Debian 12 Bookworm and have an anaconda venv of Python 3.9.19 and I installed everything in the 'requirements.txt'. The only possible issue I can think of is, my source recordings are somehow so bad that the program doesn't even know what it's doing.
Hello, I recently stumbled upon this tool as I'm trying to train Coqui on my own voice. However, I can't seem to get the program to work as intended, with the two main issues mentioned in the title.
Issue 1 - 'Immediately' Ignoring Longer Files (>5 min)
The program would split the files successfully but would output nothing in the zip, I had to modify silence threshold for it to work, as the original configuration would literally immediately ignore it
Issue 2 - Shorter files are not being split in the zip
I've tried uploading several smaller files and they seem to get split, transcribed. However, the zip output is just my original wav file and a single line of transcription, just for that file.
Edit: By 'successful split', I mean it get outputted to 'wavs_split_final' folder.
I have no idea what I'm doing wrong, I'm running Debian 12 Bookworm and have an anaconda venv of Python 3.9.19 and I installed everything in the 'requirements.txt'. The only possible issue I can think of is, my source recordings are somehow so bad that the program doesn't even know what it's doing.