基于联合语音出现概率的先验信噪比估计算法

先验信噪比是语音增强的关键参数。该文分析了几种典型的先验信噪比估计算法,并得到这几种算法的统一形式,最后提出了基于联合语音出现概率的先验信噪比估计算法。测试结果表明,该算法在不引入音乐噪声的同时,平均段信噪比提高和平均对数谱距离等客观评价指标,都优于其它算法。

A Modified a Priori SNR Estimator Based on the United Speech Presence Probabilities

The a priori Signal-to-Noise Ratio (SNR) is the dominant parameter in noise reduction techniques. This paper analyses several typical estimate approaches of the a priori SNR firstly, and then expresses them in the same form, finally presents an improved estimate of the a priori SNR, which is based on the united speech presence probabilities. Simulation results verify that the performance of the proposed estimator is better than those conventional estimators in terms of the segmental SNR and the log-spectral distance.