基于时频结合的背景噪声下语音增强方法

Method for Speech Enhancement in Noisey Environment Based on Time-Frequency Joint

本文在研究基于改进谱减法的基础之上 ,提出了在频域利用谱减法去除宽带加性噪声 ,在时域利用短时过零率和短时能量组合而成的加权函数去除谱减法后产生的“音乐噪声”的方法 .实验表明

As speech recognition has moved from the laboratory to the real field , recognition scores drop significantly. Denoising and robust speech recognition are of critical importance in practical deployment of speech technology. The enhancement of noisy speech has gained an increasing interest in recent years. In this paper, we present a new speech enhancement method, which uses spectral and time domain processing and achieves a trade off between effective and low computational load for real time operations. In the spectral domain, a improved spectral subtraction is used to reduce the effect of additive broadband noise on speech. In the time domain, a weighting function is used to reduce a residual 'musical noise'. Experimental results show this method can offer better performance.