A fast EM algorithm for Gaussian model-based source separation

Joachim Thiemann (1, 2) , Emmanuel Vincent (3)

1 PANAMA - Parcimonie et Nouveaux Algorithmes pour le Signal et la Modélisation Audio
2 METISS - Speech and sound data modeling and processing
3 PAROLE - Analysis, perception and recognition of speech

We consider the FASST framework for audio source separation, which models the sources by full-rank spatial covariance matrices and multilevel nonnegative matrix factorization (NMF) spectra. The computational cost of the expectation-maximization (EM) algorithm in [1] greatly increases with the number of channels. We present alternative EM updates using discrete hidden variables which exhibit a smaller cost. We evaluate the results on mixtures of speech and real-world environmental noise taken from our DEMAND database. The proposed algorithm is several orders of magnitude faster and it provides better separation quality for two-channel mixtures in low input signal-to-noise ratio (iSNR) conditions.

Mots clés: audio source separation; FASST; EM algorithm; binary masking; DEMAND

Domaines: Traitement du signal et de l'image [eess.SP]; Traitement du signal et de l'image [eess.SP]

Joachim Thiemann, Emmanuel Vincent. A fast EM algorithm for Gaussian model-based source separation. EUSIPCO - 21st European Signal Processing Conference - 2013, Sep 2013, Marrakech, Morocco. ⟨hal-00840366⟩