A Speech Enhancement System Using Binary Mask Approach and Spectral Subtraction Method

In this paper, we proposed a speech enhancement system by combining the binary time-frequency mask and spectral subtraction method. It is constructed in Short Time Fourier Transform (STFT) domain with feature extraction assistance. The binary mask approach and spectral subtraction techniques are also applied. Analysis is done on noisy speech signal corrupted by factory noise at -3dB, 0dB and 3dB SNR levels. Simulation and results are performed in MATLAB 7.14 (R2012a). With the output SNR (Signal to Noise Ratio) analysis, we have obtained promising results in real experimental environment.