VERSAMUS project
Music plays a major role in everyday use of digital media contents. Companies and users are waiting for smart content creation and distribution functionalities, such as music classification, search by similarity, summarization, chord transcription, remixing and automatic accompaniment.
So far, research efforts have focused on the development of specific algorithms and corpora for each functionality based on low-level sound features characterizing sound as a whole. Yet, music generally results from the superposition of heterogeneous sound components (e.g. voices, pitched musical instruments, drums, sound samples) carrying interdependent features at several levels (e.g. music genre, singer identity, melody, lyrics, voice signal). Integrated music representations combining all feature levels would make it possible to address all of the above functionalities with increased accuracy as well as to visualize and interact with the content in a musically relevant manner.
The VERSAMUS project has been set up in 2010 as a joint research effort between the METISS team at the Institut National de Recherche en Informatique et en Automatique (INRIA, the French National Research Institute for Research in Computer Science and Control) in Rennes, France, and the Laboratory #1 at the Graduate School of Information Science and Technology, The University of Tokyo. The aim of this project is to investigate, design and validate such representations in the framework of Bayesian data analysis, which provides a rigorous way of combining separate feature models in a modular fashion. Tasks to be addressed include the design of a versatile model structure, of a library of feature models and of efficient algorithms for parameter inference and model selection. Efforts will also be dedicated towards the development of a shared modular software platform and a shared corpus of multi-feature annotated music which will be reusable by both partners in the future and eventually disseminated.
Workshop
The goal of The Second VERSAMUS Workshop on Multiresolution and Multilayer Music Processing is to present and discuss the work done by the project members from both research groups, but it will also be open for anyone who wish to participate.
If you wish to participate, please kindly send an email to visit@ at our server hil.t.u-tokyo.ac.jp.
Details
Cost: free
Time: Monday, June 20, 2011
Place: Seminar Room A&D on 3rd Floor of Engineering Building 6, Hongo Campus, Tokyo, Japan (map in Japanese, map in English).
Access: map in Japanese, map in English.
Programme
| Project member presentations 1 | 10:00 – 11:00 |
| Short break | 11:00 – 11:20 |
| Project member presentations 2 | 11:20 – 12:20 |
| Lunch break | 12:20 – 13:30 |
| Project member presentations 3 | 13:30 – 15:00 |
| Short break | 15:00 – 15:20 |
| Sagayama laboratory member presentations 1 | 15:20 – 16:00 |
| Short break | 16:00 – 16:20 |
| Sagayama laboratory member presentations 2 | 16:20 – 17:00 |
| Conclusion and general discussion | 17:00 – |
Project member presentations
Each presentation is 20 min + 10 min for questions.
1. Emmanuel Vincent 10:00 – 10:30
Overview of the Versamus project
2. Satoru Fukayama 10:30 – 11:00
Highly dimensional dependency analysis of musical components
3. Stanisław A. Raczyński 11:20 – 11:50
Dynamic Bayesian Networks for polyphonic symbolic pitch modeling
4. Nobutaka Ono (for Jun Wu) 11:50 – 12:20
Polyphonic pitch estimation by joint modeling of sustained and attack sounds
5. Nobutaka Ito 13:30 – 14:00
Robust sound source localization based on trace norm minimization
6. Ngoc Q. K. Duong 14:00 – 14:30
Multichannel harmonic and percussive component separation by joint modeling of spatial and spectral continuity
7. Hideyuki Tachibana 14:30 – 15:00
Multi-stage HPSS applied to new signal characterization and source separation
Sagayama laboratory member presentations
Presentations about research related to the project. Each presentation is 15 min + 5 min for questions.
1. Yuu Mizuno 15:20 – 15:40
Real-time modification of time scale, pitch and timbre by reconstructing phase from the modified STFT spectrogram of acoustic signals
2. Hirokazu Kameoka (for Masahiro Nakano) 15:40 – 16:00
Infinite-state spectrum model for music signal analysis
3. Kosuke Suzuki 16:20 – 16:40
Real-time audio to score alignment with online dynamic time warping
4. Kazuki Ochiai 16:40 – 17:00
Concurrent NMF of multi-resolution spectrograms for multipitch analysis of music signals