Subject: [linux-audio-dev] MARSYAS framework for Computer Audition
From: George Tzanetakis (gtzan_AT_cs.princeton.edu)
Date: Sun Dec 09 2001 - 01:24:03 EET
Dear friends,
although I have not posted any message, I have been reading this list
for some time.I am graduate student finishing my PhD degree in Computer
Science
in Princeton University under the guidance of Prof. Perry Cook.
For the last few years I have been conducting research in
Computer Audition (examples: automatic music segmentation,
musical genre classification, feature extraction, music/speech
classification, similarity retrieval etc)
essentially developing and combining techniques from
Signal Processing and Machine Learning to analyze audio and especially
music.
Mainly for my own research I have developed a software
framework in C++ called MARSYAS that contains all my proposed
algorithms as well as many of the existing computer audition
algorithms. A small-scale graphical user interface written
in JAVA can also be used as a prototype computer audition enhanced
audio editor. Using MARSYAS audio files can be automatically classified
and new classification schemes can easily be created and evaluated.
MARSYAS has also been used by undergraduate and graduate students
in Princeton doing computer audition projects.
I believe it should be straightforward to use MARSYAS
for example to write plugins that perform musical genre
classification or similarity retrieval
for the audio systems described in this list.
Therefore I thought it might be of interest to people in this list
to take a look at my software and possibly use it.
MARSYAS is available as free software under the GPL
from http://www.cs.princeton.edu/~gtzan/marsyas.html
and should compile easily under Linux.
Alhtough I have very little time since this is my last year
I would be happy to help and collaborate with anyone who might find
MARSYAS
useful. Also please do not hesitate to send any feedback (good or bad)
to
gtzan_AT_cs.princeton.edu. Please bear in mind that this software
was created for my own research and although I have made an effort
to make it clean and well-documented a lot of work remains to be
done in this area.
Thank you for your time,
George Tzanetakis
This archive was generated by hypermail 2b28 : Sun Dec 09 2001 - 01:24:03 EET