An improved spectral subtraction algorithm applied to speech enhancement in the cochlear implant / 生物医学工程学杂志

ABSTRACT

The performance of the cochlear implant (CI) deteriorates rapidly in the presence of background noise, and the conventional spectral subtraction speech enhancement method used in the CI can not remove the non-stationary noise well. In order to improve the audio perception ability of the CI users, an improved spectral subtraction method is proposed, which divides the power spectrum of the noisy speech according to the Bark bands and estimates the noise by using the minima controlled recursive averaging method in every individual band when the speech frame is present. Then the adaptive spectral subtraction parameters are used in the algorithm. The proposed spectral subtraction method has been applied in CI, and the simulation result shows that the proposed algorithm may extract the speech features more correctly and the CI synthesized sound is of higher quality.