Single channel speech enhancement using temporal masking

Abstract

A novel speech enhancement technique is presented based on the temporal masking properties of the human auditory system. The input signal is divided into a number of sub-bands that are individually and adaptively weighted in the time domain according to a short term SNR estimate in each sub-band. The short time SNR estimate is calculated using the temporal masking thresholds in each sub-band. Objective measures and informal listening tests demonstrated significant improvements over other methods when tested with speech signals corrupted by car noise for SNR values of 0, 10, and 20 dB