Alessio Bazzica

Alessio Bazzica's picture

Alessio Bazzica

Blog
 
View Alessio Bazzica's profile on LinkedIn
 
Electrical Engineering, Mathematics and Computer Science
Intelligent Systems
Mekelweg 4
2628 CD Delft
The Netherlands

Office: HB11.320
In symphonic music, many musicians perform at the same time. Can we make sense of the scene and analyze the individual contributions of each performing musician?

Most of the problems in Music Information Retrieval (Music IR) are solved by analysis of the audio channel. However, when focusing on the contributions of isolated musicians or instrumental sections, the analysis of audio recordings may face critical challenges like, for instance, those related to source separation, especially in the case of large ensembles where the sound produced by different instruments overlaps both in time and frequency.
Alternatively, one can exploit symbolic information but it is only useful when synchronized to the audio-visual recording of the analyzed performance. While crawling symbolic scores (e.g. MIDI files) is often possible, reliably synchronizing them can be challenging especially in the presence of structural changes (e.g. omissions, repetitions).

We investigate how to possibly exploit the visual channel of audio-visual recordings to improve existing Music IR algorithms and design new annotation and visualization systems for symphonic music videos.

On http://youtu.be/6Oj70tvqo9c, you can watch a video in which we demonstrate ongoing work on visual analysis of symphonic concert recordings. We also show some novel visualizations and possible applications built on top of our video analysis method's output also combining it to that of performance-to-score synchronization and that of isolated sound source separation.

Research areas:

Publication list
Google Scholar

Datasets:

  • Quasi-Static Scenes (QSS) video dataset
  • Clarinetist 4 Science (C4S) audiovisual dataset
CV essentials:
Google+ PhD Intern (Nov 2015 - Feb 2016)
Google, Stockholm (Sweden)
Visiting Researcher (Jun-Jul 2014)
UPF Music Technology Group, Barcelona (Spain)
MEng Computer Engineering Student (2008-2012)
University of Florence (Italy)
Visiting Student (Feb-Jul 2012)
TU Delft, Delft Multimedia Information Retrieval lab (The Netherlands)
BEng Graduating Student (Jan-Sep 2008)
Media Integration and Communication Centre of excellence, University of Florence (Italy)
BEng Computer Engineering Student (2002-2008)
University of Florence (Italy)