Single-Channel Speech Separation Using Phase-Based Methods

Yun-Kyung Lee

doi:10.1109/TCE.2010.5681127

2010-11-01

PRE

Yun-Kyung Lee *

In Lee

Oh‐Wook Kwon

DOI：10.1109/TCE.2010.5681127

原文链接

原文求助

摘要

En 中文

This paper addresses the problem of single-channel speech separation to extract and enhance the desired speech signal from mixed speech signals. We propose a new speech separation algorithm by utilizing both magnitude and phase information, which can be applied to multimedia mobile communication and navigation systems. Conventionally, phase information has been neglected in speech signal processing. However, in the proposed method, we formulate a probabilistic phase-based speech estimator based on zero-phase models to improve the speech separation performance. In the speech separation experiments, the proposed method is shown to improve speaker-to-interference ratio (SIR) by 2.2 dB compared to the system using magnitude models only. When only phase-based speech estimator is used for speech separation, the SIR was improved by 0.8 dB. This result justify that the proposed phase-based speech estimator achieves significant SIR improvement compared with the previous magnitude-based method(1).

Keyword:

phase modeling

soft mask

speech separation

speech enhancement