An EM Algorithm for Joint Source Separation and Diarisation of Multichannel Convolutive Speech Mixtures

Dionyssos Kounades-Bastian (1) , Laurent Girin (2, 1) , Xavier Alameda-Pineda (3, 1) , Sharon Gannot (4) , Radu Horaud (1)

1 PERCEPTION - Interpretation and Modelling of Images and Videos
2 GIPSA-CRISSP - GIPSA - Cognitive Robotics, Interactive Systems, & Speech Processing
3 UNITN - Università degli Studi di Trento = University of Trento
4 Bar-Ilan University [Israël]

We present a probabilistic model for joint source separation and diarisation of multichannel convolutive speech mixtures. We build upon the framework of local Gaussian model (LGM) with non-negative matrix factorization (NMF). The diarisa-tion is introduced as a temporal labeling of each source in the mix as active or inactive at the short-term frame level. We devise an EM algorithm in which the source separation process is aided by the diarisation state, since the latter indicates the sources actually present in the mixture. The diarisation state is tracked with a Hidden Markov Model (HMM) with emission probabilities calculated from the estimated source signals. The proposed EM has separation performance comparable with a state-of-the-art LGM NMF method, while out-performing a state-of-the-art speaker diarisation pipeline.

Mots clés: speaker diarisation; local Gaussian model; Audio source separation

Domaines: Son [cs.SD]; Traitement du signal et de l'image [eess.SP]; Apprentissage [cs.LG]

Dionyssos Kounades-Bastian, Laurent Girin, Xavier Alameda-Pineda, Sharon Gannot, Radu Horaud. An EM Algorithm for Joint Source Separation and Diarisation of Multichannel Convolutive Speech Mixtures. ICASSP 2017 - IEEE International Conference on Acoustics, Speech and Signal Processing, Mar 2017, New Orleans, United States. pp.16-20, ⟨10.1109/ICASSP.2017.7951789⟩. ⟨hal-01430761⟩