A Speech Enhancement Algorithm combining Spectral Subtraction and Wavelet Transform
Yi Yang, Peipei Liu, Huili Zhou, Ye Tian · 2021 IEEE 4th International Conference on Automation, Electronics and Electrical Engineering (AUTEEE) · 2021
Spectral subtraction is widely used in the research of speech enhancement because of its small calculation amount and fast processing speed. However, traditional spectral subtraction has a large amount of residual music noise in the process of speech noise reduction. In order to solve this problem, a speech enhancement algorithm based on improved spectral subtraction and improved threshold wavelet transform is proposed. The method first uses spectral subtraction to denoise the noisy speech signal, and then the processed signal is subjected to wavelet transform perform voice enhancement. By using MATLAB software to simulate the system, the experimental results show that compared with the traditional spectrum subtraction method, the improved method can effectively suppress the music noise generated by the single spectrum subtraction method, improve the output signal-to-noise ratio and recognizability of the speech signal, and enhance the speech The effect is remarkable.