Team METISS

Members
Overall Objectives
Scientific Foundations
Application Domains
Software
New Results
Contracts and Grants with Industry
Other Grants and Activities
Dissemination
Bibliography

Section: New Results

Keywords : music, pitch transcription, language model, n-gram.

Content description of music signals

Multi-pitch signal modeling

Participant : Emmanuel Vincent.

Main collaboration: P. Leveau, N. Bertin (Telecom ParisTech)

Music involves several levels of information, from the acoustic signal up to cognitive quantities such as composer style or key, through mid-level quantities such as a musical score or a sequence of chords. The dependencies between mid-level and lower- or higher-level information can be represented through acoustic models anD lanwuge models, respectively. Given some limitations of existing acoustic models, including our previous time-domain models [19] , [20] , we proposed a frequency-domain acoustic model that exploits the timbre of each instrument to increase the accuracy of the inferred musical score without relying on separate training data. This model represents an input short-term magnitude spectrum as a linear combination of magnitude spectra corresponding to different pitches, which are adapted to the input under harmonicity constraints [37] .

Music language modeling

Participants : Emmanuel Vincent, Frédéric Bimbot.

Main collaboration: Ricardo Scholz (internship student)

We started working on the modeling of music as a language by studying N-gram models of chord sequences. We investigated various chord labelling schemes and various model smoothing techniques originally designed for spoken language processing. While state-of-the-art models consider N=2, we showed that more accurate models with N > 2 could be learned from a limited set of data [52] .


previous
next

Logo Inria