[opensource] [AI] AI-Group Talk (this Friday / 3:30 pm / DL357)

Michael S. Yanovich yanovich.1 at osu.edu
Thu Nov 4 19:14:30 EDT 2010

I think this is interesting.

-------- Original Message --------
Subject: [AI] AI-Group Talk (this Friday / 3:30 pm / DL357)
Date: Mon, 01 Nov 2010 15:05:59 -0400
From: Ryan <ryanbuaa at gmail.com>
To: ai at cse.ohio-state.edu, Leon Hsu <leon at mirlab.org>

Speaker: Chao-Ling Hsu
Time: Friday, Nov. 5th, 3:30 pm
Location: Dreese Labs 357

Title: Singing Pitch Extraction and Voice Separation from Music

Computational auditory scene analysis (CASA) has been shown very effective
in separating speech from various kinds of background noise for monaural
mixtures. Target pitch detection is the key phase of the current CASA
systems and dominates the performance of the system. Unfortunately, it is
difficult to detect the target pitch robustly especially for the mixtures
with non-stationary and strong harmonicity noise such as music. Pitch from
this type of noise may mislead the target pitch detection algorithm
seriously. This paper proposes an extended tandem algorithm that estimates
the singing pitches and separates the singing voice jointly and iteratively.
Rough pitches are first estimated and are used to separate the target
singing by considering harmonicity and temporal continuity. The separated
singing voice and estimated pitches are then used to improve each other
iteratively. To improve the performance of the tandem algorithm for dealing
with musical recordings, we propose a trend estimation algorithm to detect
the pitch ranges of a singing voice in each time frame. The detected trend
substantially reduces the difficulty of singing pitch detection by reducing
a large number of wrong pitch candidates either produced by musical
instruments or the overtones of the singing voice. This algorithm greatly
increases the performance of the system by giving a much more accurate pitch
estimation. Systematic evaluation shows that the extended tandem algorithm
outperforms the previous systems for either pitch extraction or singing
voice separation.

Chao-Ling Hsu was born in Taiwan, R.O.C.. He received his B.S. degree in
computer science and information engineering from Chung-Hua University,
Hsinchu, Taiwan, in 2003. He is currently pursuing a Ph.D. degree in
computer science at National Tsing Hua University, Hsinchu, Taiwan and now a
visiting student in The Ohio State University, Columbus, US, from Feb. 2010
to now. His research interests include melody recognition, computational
auditory scene analysis, and music signal processing.

**Refreshments will be served.**

Yanzhang(Ryan) He

Department of Computer Science and Engineering
the Ohio State University - Columbus

Mobile: 1-614-284-6980

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 899 bytes
Desc: OpenPGP digital signature
Url : http://mail.cse.ohio-state.edu/pipermail/opensource/attachments/20101104/c70225e6/signature.bin

More information about the Opensource mailing list