Speech Enhancement Using a Novel Spectral Subtraction Method for Seashore Noise

Many factors including noise are there for degradation of the voice signal that needs to be enhanced. In this paper, we have taken an attempt to improve the speech quality. The averaging spectral subtraction (SS) method is modified with the Welch method. In first phase, the Bartlett method is used and in the second phase, the Welch method is used as the novelty of the work. The method is based on the spectrum dependent gain function and the noise is estimated using Welch method. The overlapping windowed frames and the adaptive averaging of the gain function result better enhancement. The results obtained using the proposed method are shown in the result section. The comparison results prove the proposed algorithm is better for tidy noisy signals. For quality measure of the signals, signal-to-noise ratio (SNR), segmental SNR (SegSNR), and mean opinion score (MOS) are evaluated.