|
Underdetermined source separation by ICA and homomorphic signal processing
- Author(s)
-
- Stefan Winter, (NTT Communication Science Laboratories, NTT Corporation)
- Walter Kellermann, (Department of Multimedia Communication and Signal Processing, University Erlangen-Nuremberg)
- Hiroshi Sawada, (NTT Communication Science Laboratories, NTT Corporation)
- Shoji Makino, (NTT Communication Science Laboratories, NTT Corporation)
- Topics
-
- Sound enhancement and sound separation
- Microphone arrays and array signal processing
|
Get the paper in PDF format
Acrobat Reader
(version 5 minimum)
is necessary to read this document.
|
Abstract
Nearly all approaches for underdetermined blind source separation (BSS) assume independent and identically distributed (i.i.d.) sources. They completely ignore the temporal structure of colored sources such as speech signals. Instead, we propose a multivariate model based on a multivariate Gaussian distribution that is then used to determine an unmixing matrix for underdetermined BSS. Based on parameterization by cepstral coefficients we present a novel ICA-based cost function for estimating the speech-related parameters of the unmixing matrix. Experimental results support the proposed approach.
|
