Speech Enhancement using Spectral Subtraction in EMD domain under Strong Noise Environment
Zhang Xiaolin
1
张晓林(1971-),男,副教授,主要研究方向为宽带数字通信系统、通信信号检测与识别
Zhen Haochuan
1
甄浩川(1986-)男,硕士研究生,主要研究方向为语音增强
1、College of Information and Communication Engineering, Harbin Engineering University, Harbin 150001
Abstract:Traditional spectral subtraction (SS) performs badly in strong noise environment. In this paper, we discuss the application of Empirical Mode Decomposition (EMD) in speech signal analysis, and then propose a novel approach to speech enhancement using SS in EMD domain (SS-EMD). Based on EMD, the noise reduction scheme is fully data-driven. The basic principle of the method is the signal reconstruction with IMFs, which use SS each. The improvement of SS here is the noise estimation using voice activity detection (VAD).It makes denoising more accurate and reasonable. A comparison between SS and SS-EMD is performed by using 'white noise' which is from Noisex92 database. The results demonstrate that the new method achieves better speech quality in strong noise environment.