Speech Enhancement Algorithm of Binary Mask Estimation Based on Noise Spectrum Constraints

In order to improve speech intelligibility with low signal-to-noise ratio, a speech enhancement of the binary mask estimation based on noise spectrum constraints is proposed. We first analyzed the impact of priori signal-to-noise ratio over-estimation on the noise spectrum estimation function with low signal-to-noise ratio. Then we corrected the priori signal-to-noise ratio and noise spectrum estimation function. At last, we used the corrected value of noise spectrum estimation function and priori signal-to-noise ratio to judge the time-frequency units where the noise spectrums were under-estimated, estimating the binary mask value, and then made a constraints on the enhanced speech time-frequency units. Simulation results show that under the several common background noise with low signal-to-noise ratio, the proposed approach is more excellent and can improve the speech intelligibility effectively compared with traditional wiener filtering method.