Speech source separation using the IVA algorithm with multivariate mixed super Gaussian student's T source prior in real room environment

Rafique, W; Naqvi, SM; Chambers, JA

doi:10.1049/cp.2015.1789

Speech source separation using the IVA algorithm with multivariate mixed super Gaussian student's T source prior in real room environment

Lookup NU author(s): Waqas Rafique, Dr Mohsen Naqvi, Professor Jonathon Chambers

Downloads

Full text for this publication is not currently held within this repository. Alternative links are provided below where available.

Abstract

Independent vector analysis (IVA) can theoretically avoid the permutation problem in frequency domain blind source separation by using a multivariate source prior to retain the dependency between different frequency bins of each source. In this paper a mixture of multivariate super Gaussian distribution and multivariate Student's t distribution is adopted as a source prior for the IVA algorithm. The Student's t distribution due to its heavy tail nature can better model the high amplitude information in the frequency bins and at the same time a dependent super Gaussian distribution can be adopted to model other information in the frequency bins. The weight of both distributions is varied in the source prior mixture and their separation performance is tested over different scenarios by using the image and binaural room impulse responses (BRIRs). Detailed simulation results confirms that the IVA algorithm with the proposed mixed multivariate source prior can consistently achieve better separation performance.