On linear and mixmax interaction models for single channel source separation

Robert Peharz, Franz Pernkopf

Onderzoeksoutput: Hoofdstuk in Boek/Rapport/CongresprocedureConferentiebijdrageAcademicpeer review

5 Citaten (Scopus)

Samenvatting

For model-based single channel source separation, one typically assumes a linear interaction model, i.e. that the mixture magnitude spectrogram is the sum of the individual source magnitude spectrograms. In the log-domain, the MIXMAX interaction model is the corresponding approximation for the linear model. Hence, one would expect similar performance for both approaches. However, in this paper we empirically show that this is not the case for vector-quantizer-based (VQ) single channel source separation. We propose factorial linear-VQ, the linear counterpart to factorial max-VQ, and compare the two methods in systematic source separation experiments. Linear-VQ performs significantly better than max-VQ for comparable code-book sizes and behaves more robustly in the presence of additive white noise. Furthermore, we compare resynthesis properties of binary and continuous time-frequency masks. While binary masks achieve a higher interference suppression, the use of continuous masks results in a consistently better signal quality.

Originele taal-2Engels
Titel2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012 - Proceedings
UitgeverijIEEE Press
Pagina's249-252
Aantal pagina's4
ISBN van geprinte versie9781467300469
DOI's
StatusGepubliceerd - 23 okt 2012
Extern gepubliceerdJa
Evenement2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012 - Kyoto, Japan
Duur: 25 mrt 201230 mrt 2012

Congres

Congres2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012
LandJapan
StadKyoto
Periode25/03/1230/03/12

Vingerafdruk Duik in de onderzoeksthema's van 'On linear and mixmax interaction models for single channel source separation'. Samen vormen ze een unieke vingerafdruk.

Citeer dit