zbMATH — the first resource for mathematics

Discovering convolutive speech phones using sparseness and non-negativity. (English) Zbl 1173.94378
Davies, Mike E. (ed.) et al., Independent component analysis and signal separation. 7th international conference, ICA 2007, London, UK, September 9–12, 2007. Proceedings. Berlin: Springer (ISBN 978-3-540-74493-1/pbk). Lecture Notes in Computer Science 4666, 520-527 (2007).
Summary: Discovering a representation that allows auditory data to be parsimoniously represented is useful for many machine learning and signal processing tasks. Such a representation can be constructed by Non-negative Matrix Factorisation (NMF), which is a method for finding parts-based representations of non-negative data. Here, we present a convolutive NMF algorithm that includes a sparseness constraint on the activations and has multiplicative updates. In combination with a spectral magnitude transform of speech, this method extracts speech phones that exhibit sparse activation patterns, which we use in a supervised separation scheme for monophonic mixtures.
For the entire collection see [Zbl 1129.94002].
94A12 Signal theory (characterization, reconstruction, filtering, etc.)
65F30 Other matrix algorithms (MSC2010)
BSS Eval
Full Text: DOI