Non - Audible Murmur (NAM) Voice Conversion by Wavelet Transform

In this paper, a statistical approach is presented to enhance the body-conducted unvoiced speech for silent speech communication. Body conducted unvoiced speech is difficult to use in human-to-human speech communication because it sounds unnatural and less intelligible owing to the acoustic change caused by body conduction. To address this issue, voice conversion (VC) method from NAM to normal speech (NAM-to-Speech) using Wavelet transform is proposed, where the acoustic features of body-conducted unvoiced speech are converted into those of natural voices. Wavelet analysis is capable to extract the data from the murmur and then it passes the data into a classifier for the recognition of isolated words. Simulation results show that NAM-to-Speech effectively improves intelligibility.