Non-negative Hidden Markov Modeling of Audio with Application to Source Separation

Submitted by gautham on Sat, 10/02/2010 - 11:03am

Title	Non-negative Hidden Markov Modeling of Audio with Application to Source Separation
Publication Type	Conference Paper
Year of Publication	2010
Authors	Mysore, G. J., P. Smaragdis, and B. Raj
Conference Name	International Conference on Latent Variable Analysis and Signal Separation (LVA / ICA)
Date Published	09/2010
Conference Location	St. Malo, France
Abstract	In recent years, there has been a great deal of work in modeling audio using non-negative matrix factorization and its probabilistic counterparts as they yield rich models that are very useful for source separation and automatic music transcription. Given a sound source, these algorithms learn a dictionary of spectral vectors to best explain it. This dictionary is however learned in a manner that disregards a very important aspect of sound, its temporal structure. We propose a novel algorithm, the non-negative hidden Markov model (N-HMM), that extends the aforementioned models by jointly learning several small spectral dictionaries as well as a Markov chain that describes the structure of changes between these dictionaries. We also extend this algorithm to the non-negative factorial hidden Markov model (N-FHMM) to model sound mixtures, and demonstrate that it yields superior performance in single channel source separation tasks.
URL	https://ccrma.stanford.edu/~gautham/Site/NFHMM.html

Search this site:

Fall Courses at CCRMA

Music 15N The Aesthetics of Data
Music 101 Introduction to Creating Electronic Sounds
Music 154A (ARTSTUDI 131) Sound Art I
Music 192A Foundations of Sound Recording Technology
Music 201 CCRMA Colloquium
Music 203 Audiovisual Performance
Music 220A Foundations of Computer-Generated Sound
Music 220D Research Topics in Computer Music
Music 251 Psychophysics and Music Cognition
Music 256A Music, Computing, and Design I: Software Paradigms for Computer Music
Music 285 Intermedia Lab
Music 319 Research Seminar on Computational Models of Sound Perception
Music 320A Introduction to Audio Signal Processing Part I: Spectrum Analysis
Music 351A Research Seminar in Music Perception and Cognition I
Music 451A Auditory EEG Research I

Main menu

Secondary menu

Non-negative Hidden Markov Modeling of Audio with Application to Source Separation

Search this site:

Fall Courses at CCRMA