Auxiliary function based IVA using a source prior exploiting fourth order relationships

Yanfeng Liang (1) , Jack Harris (1, 2) , Gaojie Chen (1) , Syed Mohsen Naqvi (1) , Christian Jutten (2) , Jonathon Chambers (1)

1 School of Electronic, Electrical and Systems Engineering
2 GIPSA-VIBS - GIPSA - Vision and Brain Signal Processing

Independent vector analysis (IVA) can theoretically avoid the permutation ambiguity present in frequency domain independent component analysis by using a multivariate source prior to retain the dependency between different frequency bins of each source. The auxiliary function based independent vector analysis (AuxIVA) is a stable and fast update IVA algorithm which includes no tuning parameters. In this paper, a particular multivariate generalized Gaussian distribution source prior is therefore adopted to derive the AuxIVA algorithm which can exploit fourth order relationships to better preserve the dependency between different frequency bins of speech signals. Experimental results confirm the improved separation performance achieved by using the proposed algorithm.

Mots clés: AuxIVA; multivariate generalized Gaussian distribution; fourth order relationships

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

Yanfeng Liang, Jack Harris, Gaojie Chen, Syed Mohsen Naqvi, Christian Jutten, et al.. Auxiliary function based IVA using a source prior exploiting fourth order relationships. EUSIPCO 2013 - 21th European Signal Processing Conference, Sep 2013, Marrakech, Morocco. pp.EUSIPCO 2013 1569746061. ⟨hal-00867131⟩