Speech Enhancement Under a Combined Stochastic-Deterministic Model

Author

Hendriks, Richard C. ; Heusdens, Richard ; Jensen, Jesper

Author_Institution

Dept. of Mediamatics, Delft Univ. of Technol.

Volume

1

fYear

2006

fDate

14-19 May 2006

Abstract

Most DFT domain based enhancement methods rely on stochastic models to derive clean speech estimators. In this paper we investigate the use of a deterministic speech model and present an MMSE estimator under a combined stochastic-deterministic speech model. Experimental results show an increase in segmental SNR of 1.18 dB, compared to the use of a stochastic model alone. Furthermore, PESQ evaluations lead to an increase of 0.3 on the MOS scale. Listening tests show a preference for the proposed MMSE estimator under combined stochastic-deterministic speech model

Keywords

discrete Fourier transforms; least mean squares methods; speech enhancement; stochastic processes; DFT domain based enhancement methods; MMSE estimator; MOS scale; PESQ evaluations; clean speech estimators; combined stochastic-deterministic speech model; segmental SNR; speech enhancement; Acoustic noise; Additive noise; Discrete Fourier transforms; Frequency; Gaussian distribution; Gaussian noise; Noise level; Speech enhancement; Stochastic processes; Stochastic resonance;

fLanguage

English

Publisher

ieee

Conference_Titel

Acoustics, Speech and Signal Processing, 2006. ICASSP 2006 Proceedings. 2006 IEEE International Conference on

Conference_Location

Toulouse

ISSN

1520-6149

Print_ISBN

1-4244-0469-X

Type

conf

DOI

10.1109/ICASSP.2006.1660055

Filename

1660055