Training To make .wav files from mp3 run > python peoplesounds/decode.py {path to base} To run training > python peoplesounds/train.py {path to base} {path to output} I've been using "data/run.txt" for output. The code will append to the file. For training - 'y' is for voice - 'n' is no voice - 's' is to skip (half voice/half no-voice)