A novel two stage single channel speech enhancement technique

A two stage novel speech enhancement algorithm is presented in this paper using speech and noise spectral estimators with hybrid a priori SNR combined with perceptual weighting filter. The first stage is designed such that it improves the performance of Minimum Mean Square Error Short-Time Spectral Amplitude (MMSE-STSA) estimator by using hybrid a priori SNR and combining statistical estimators of the spectral magnitude of the speech and noise. A perceptual weighting filter is designed and used as a second stage to improve the intelligibility of the processed speech signal. The performance of proposed scheme is tested and results indicate the performance of the proposed algorithm is better than several other MMSE-STSA algorithms available in literature.