- Title:
- Musical Scene Analysis and MPEG-7
- Presenter:
- Thibaut Sacreste
- Date:
- July 05, 2002
- Abstract:
- Musical Scene Analysis is the segregation of a mixture of musical sounds
in relation to the different instruments composing it. Many attempts have
been made to create computational models that can perform this task
However, none of them come close to reaching the performance level of
the human ear. An overview of previous works in Musical Scene Analysis
and Auditory Scene Analysis will be given in order to identify the issues
behind their application to computational models. Then, a simple model for
Musical Scene Analysis will be presented. It uses machine learning
methods and the MPEG-7 low-level audio descriptors. Finally, results
obtained with our model will be presented, and possible improvements
discussed.
Back to:
|
