- audio cleaning
- detect non-speech vs speech
- can keras be useful for audio classification?
- spectra into random forest or other (un)supervised classifier?
- frequencies vs amplitudes - which is better?
- envelope fingerprinting or phonemes
- time autocorrelations or cross-correlations?
- cluster analysis to identify groups of speakers
- identify what makes clusters fail
- http://samcarcagno.altervista.org/blog/basic-sound-processing-python/?doing_wp_cron=1560116490.2466609477996826171875
- https://www.kaggle.com/kernels/scriptcontent/1762680/download
- https://medium.com/@ageitgey/machine-learning-is-fun-part-6-how-to-do-speech-recognition-with-deep-learning-28293c162f7a