A Single-Channel Speech Enhancement Algorithm Combined with Time-Frequency Mask

Aiming at the problem of noise overestimation when classical single-channel speech enhancement method suppresses noise, on basis time-frequency mask is put forward to improve quality and separation. First, we estimate prior signal-to-noise ratio from noisy speech, calculate finally combine synthesis enhance speech. Adaptive combines IBM IRM avoid over-suppressing noise. In this paper, processed by an improved combined with SNR. The algorithm advantages mask. experimental results indicate that significantly improves ratio, robust, easy use.