An auditory perception based improved multi-band spectral subtraction algorithm for enhancement of speech degraded by non-stationary noises

In this paper, an auditory perception based improved multi-band spectral subtraction algorithm is proposed to enhance the speech signal degraded by non-stationary or colored noises. In the proposed scheme, the whole speech spectrum is divided in different non-uniform bands (N = 6) in accordance to the critical-band rate scale and spectral subtraction is applied separately in each band. The proposed algorithm uses a new approach to estimate the noise power from each band without the need of explicit speech silence detection. The noise estimate is updated by adaptively smoothing the noisy signal power. The smoothing parameter is controlled by a linear function of a-posteriori signal-to-noise ratio (SNR). This noise estimation approach gives accurate results at low SNR and works continuously in the presence of speech. The objective measures as well as informal subjective tests demonstrate that the proposed algorithm reduces remnant noise efficiently and the enhanced speech contains minimal speech distortions with improved SNR.