Spectral Intersections for Non-Stationary Signal Separation
Venue
Proceedings of InterSpeech 2012, Portland, OR
Publication Year
2012
Authors
Trausti Kristjansson, Thad Hughes
BibTeX
Abstract
We describe a new method for non-stationary noise suppression that is simple to
implement yet has performance rivaling far more complex algorithms. Spectral
Intersections is a model based MMSE signal separation method that uses a new simple
approximation to the observation likelihood. Furthermore, Spectral Intersections
uses an efficient approximation to the expectation integral of the MMSE estimate
that could be described as unscented importance sampling. We apply the new method
to the task of separating speech mixed with music. We report results on the Google
Voice Search task where the new method provides a 7% relative reduction in WER at
10dB SNR. Interestingly, the new method provides considerably greater reduction in
average WER than the MAX method and approaches the performance of the more complex
Algonquin algorithm.
