Emmanouil Benetos

Contact Details

Title: Research Student
Tel: Internal: [13] 7480
National: 020 7882 7480
International: +44 20 7882 7480
Fax: National: 020 7882 7997
International: +44 20 7882 7997
Email: emmanouil.benetos@elec.qmul.ac.uk
Website: http://www.dcs.qmul.ac.uk/~emmanouilb/
Room: Eng. 104

Research Group: Centre for Digital Music

Supervisor: Simon Dixon

Research Topic: Automatic Transcription of Polyphonic Music

Automatic music transcription is the process of converting an audio recording into a symbolic representation using musical notation. Even for expert musicians, transcribing polyphonic pieces of music is not a trivial task, and while the problem of automatic pitch estimation for monophonic signals is considered to be a solved problem, the creation of an automated system able to transcribe polyphonic music without setting restrictions on the degree of polyphony and the instrument type still remains open.

In the past years, the problem of automatic music transcription has gained considerable research interest due to the numerous applications associated with the area, such as automatic search and annotation of musical information, interactive music systems (eg. computer participation in live human performances, score following, and rhythm tracking), as well as musicological analysis. Important subtasks for automatic music transcription include pitch estimation, onset/offset detection, loudness estimation, instrument recognition, and extraction of rhythmic information.

My main objective is the development of an automatic transcription system for western classical music that can operate with a high degree of polyphony and is not limited to pitched percussive instruments, but can accurately transcribe music produced by bowed string and wind instruments. More specifically, research will focus on the presence of harmonically related notes, which still remains an open problem in the literature, by utilizing information found in the temporal evolution of the partials, in addition to the spectral structure of each segment. To that end, techniques stemming from signal processing theory and statistical pattern recognition will be employed. An unsupervised learning procedure will be developed for modeling the spectral envelope of the partial sequence for each note, so that the system can adapt itself to variable input, without being dependent on any prior training. The system is expected to produce a piano-roll representation containing information on pitch, as well as on note onsets and offsets.

Publications

For a complete list of publications click here.