Langues

Blind source separation based on independent low-rank matrix analysis and its extension to student’s t-distribution par Daichi Kitamura, 04/09/2017

—- Title —-

Blind source separation based on independent low-rank matrix analysis and its extension to student’s t-distribution

—- Abstract —-
In this talk, we introduce a new blind source separation method called independent low-rank matrix analysis (ILRMA). ILRMA is a unified method of traditional frequency-domain independent component analysis (ICA) and nonnegative matrix factorization (NMF), where the frequency-wise demixing matrix can be estimated by capturing low-rank time-frequency structures (power spectrogram) of each source with NMF. Also, Student’s t-distribution, which includes two alpha-stable distributions as special cases, is newly employed as a super Gaussian source distribution in ILRMA.

Comments are closed.