Reduction of non-stationary noise using a non-negative latent variable decomposition

Mikkel N. Schmidt, Jan Larsen

AbstractWe present amethod for suppression of non-stationary noise
in single channel recordings of speech. Themethod is based
on a non-negative latent variable decomposition model for
the speech and noise signals, learned directly from a noisy
mixture. In non-speech regions an overcomplete basis is
learned for the noise that is then used to jointly estimate
the speech and the noise from the mixture. We compare
the method to the classical spectral subtraction approach,
where the noise spectrum is estimated as the average over
non-speech frames. The proposed method significantly outperforms
the classic approach, especially when the noise is
highly non-stationary and at low signal-to-noise ratios.
Keywordssingle channel noise reduction, latent variable, bayes, non-negative factorization
TypeMisc [Presentation]
Journal/Book/ConferenceIEEE Workshop on Machine Learning for Signal Processing 2008
Year2008    Month October
PublisherDepartment of Informatics and Mathematical Modelling
AddressRichard Petersens Plads, Building 321
NotePresented at MLSP2008 19.10.2008
Electronic version(s)[pdf]
Publication link
BibTeX data [bibtex]
IMM Group(s)Intelligent Signal Processing

Back  ::  IMM Publications