persephone
persephone copied to clipboard
If WAVs are empty, present a warning on feature extraction, skipping the file instead of crashing
This is probably a good spot for using the warnings module and logging any occurrences where this happens.
Status on this one? How is emptiness defined in this context?
Nothing has changed. Empty means there is no actual WAV data. It's just a header with duration 0 WAV.
Another related thing would be to skip utterances where the number of frames after feature extraction is less than the number of labels in the corresponding transcription, as that will break the CTC algorithm.
It's just a header with duration 0 WAV.
That should be fairly easy to check for. The utterances check is far more subtle, good catch.