Conventional post-filtering (CPF) algorithms often use a fixed filter bandwidth to estimate the auto-spectra and the cross-spectrum. This paper first studies the drawback of the CPF algorithms under the stochastic model and discusses the ways to improve the performances of the CPF algorithms. To improve noise reduction without introducing audible speech distortion, we propose a novel spectral estimator, which is based on the structure of the noise power spectral density (NPSD). The proposed spectral estimator is applied to improve the performance of the CPF. Experimental results verify that the proposed algorithm is better than the CPF algorithms in terms of the segmental signal-to-noise-ratio improvement and the noise reduction, especially the noise reduction, is about 6 dB higher than the CPF.

Authors: Wang, Jie; Zheng, Chengshi; Zhang, Chunliang; Sun, Yueyan
Affiliations: Guangzhou University, Guangzhou, China; Chinese Academy of Sciences, Beijing, China; Chinese Academy of Sciences, Shanghai, China(See document for exact affiliation information.)
AES Convention: 135 (October 2013) Paper Number: 8943
Publication Date: October 16, 2013 Import into BibTeX
Subject: Applications in Audio
Permalink: https://www.aes.org/e-lib/browse.cfm?elib=16993

Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!

This paper costs $33 for non-members and is free for AES members and E-Library subscribers.

Learn more about the AES E-Library

E-Library Location: (CD 135Papers)   /conv/135/8943.pdf

Start a discussion about this paper!