Areas of Interest

My main areas of interest are automatic music transcription and automatic analysis of harmony.

Automatic Music Transcription

Before studying computer science, I worked as musician. I used to play in bands that most of the time played covers. When doing this, my philosophy was to try to reproduce the original song as much as possible. Therefore, I used to spend most of my time trying to pick by ear every single note that was played in the original recording. For some songs, this was a very difficult task, and although after years of doing this job it became easier, it was nevertheless hard.

For that reason, I became interested in to create a computer program that would automatically transcribe music. To achieve this, some complex problems must be solved: beat quantization, timbre recognition, and melody recognition, among others. (Extraction of a score from a MIDI file is relatively easy, and is already available in commercial software). Automatic music transcription (AMT) is useful not only for professional musicians but also for amateurs. An example of general use of AMT is Query-by-humming, which consists on searching for a song not by its title, but the user-hummed version of part of the song.

Automated Analysis of Harmony

This project is an extension of the previous one. The goal is to determine the harmony (i.e. chords) of a piece of music. Automatic analysis of harmony has applications not only for professional musicians but also for amateurs. It could be used for example to extend Karaoke systems. It would let them to show not only the lyrics of a song, but also the chords, in case the singer also want to play an instrument like guitar or piano.