This is a personal projects that aims to recognize and transcribe a piano recording into a music score using only Digital Signal Processing.
My first assumption was use FFT transform, but seems that for music, a better transform is a cqt implemented in librosa
The generic problem is called Automatic Music Transcription
In the way of doing this, I found some interesting projects
- pyknon: A library to generate music, also comes with a pretty interesting book called Music for Geeks and Nerds
- pytheory: An interesting library to generate chord information