Speech enhancement using noise estimation based on dynamic quantile tracking for hearing impaired listeners

Abstract

Persons with sensorineural hearing loss experience difficulty in speech perception in noisy environments. They can benefit from signal processing for suppressing the background noise in their hearing aids. For such an application, a speech enhancement technique using spectral subtraction and noise spectrum estimation based on dynamic quantile tracking is presented. It does not involve voice activity detection or storage and sorting of past spectral samples. Enhancement of speech corrupted with different types of additive stationary and non-stationary noise showed improvement in speech quality to be equivalent to an SNR advantage of 3 - 6 dB. The technique is implemented and tested for satisfactory real-time operation, with sampling frequency of 10 kHz, on a 16-bit fixed-point DSP processor with on-chip FFT hardware.