Data pipeline to feature engineer entropy/complexity metrics of audio data. Comparative visualization notebook and excel spreadsheet included.
Initial dataset can be found here.
Lossless audio pulled from TunElf Spotify Converter.
Run aws_music.py on AWS Linux EC2 r5 Instance/S3 bucket.
See selectedSongs and visualizations for visualized data set.
TODO: Implement PyTorch Audio Feature Extraction (comparable to librosa?) / RNN on sample time series?