13th Annual Conference of the International Speech Communication Association

Portland, OR, USA
September 9-13, 2012

Phase Estimation for Signal Reconstruction in Single-channel Source Separation

Pejman Mowlaee (1), Rahim Saeidi (2), Rainer Martin (1)

(1) Institute of Communication Acoustics (IKA), Ruhr-Universität Bochum, Germany
(2) Centre for Language and Speech Technology, Radboud University Nijmegen, The Netherlands

Single-channel speech separation algorithms frequently ignore the issue of accurate phase estimation while reconstructing the enhanced signal. Instead, they directly employ the mixed-signal phase for signal reconstruction which leads to undesired traces of the interfering source in the target signal. In this paper, as- suming a given knowledge of signal spectrum amplitude, we present a solution to estimate the phase information for signal reconstruction of the sources from a single-channel mixture ob- servation. We first investigate the effectiveness of the proposed phase estimation method employing known magnitude spectra of sources as an ideal case. We further relax the ideal signal spectra assumption by perturbing the clean signal spectra via Gaussian noise. The results show that for both scenarios, ideal and noisy magnitude signal spectra, the proposed phase estima- tion approach offers improved signal reconstruction accuracy, segmental SNR and PESQ compared to benchmark methods, and those neglecting the phase information.

Index Terms: Phase estimation, signal reconstruction, single-channel speech separation

Full Paper

Bibliographic reference.  Mowlaee, Pejman / Saeidi, Rahim / Martin, Rainer (2012): "Phase estimation for signal reconstruction in single-channel source separation", In INTERSPEECH-2012, 1548-1551.