WAVELET THRESHOLDING TECHNIQUES IN SPEECH SIGNAL DENOISING

摘要

Voice produced by laryngectomy subjects implanted with voice prosthesis is considered in this study. Hard and soft wavelet thresholding techniques and Neighshrink denoising techniques are implemented. The performance of denoising is evaluated through objective measures. The quantitative results obtained are compared to assess the best performing technique. The performance of Neighshrink is observed to be significantly better than the hard and soft thresholding techniques. Neighshrink is able to mitigate noise and enhance speech quality without introducing speech distortion.