Machine Learning Methods for Music Transcription

Music transcription is the transformation of musical sound signals into human-readable form. Some trained people could achieve this task, with the help of their education and the human intuition. But it is very challenging to develop an automated transcriber, because we need to equip machines with this intuition. We do not realize it often, but there is great mathematical harmony in music. So if we could figure out the aspect of mathematics thoroughly, it would help us develop new technologies for the analysis and synthesis of music. In this project, I made research about properties of musical sound signals, and implemented machine learning methods for analyzing and transcribing them.