![]() | Only 14 pages are availabe for public view |
Abstract Dysarthria is a motor speech disorder that is often associated with irregular phonation (e.g. vocal fry) and amplitude, in coordination of articulators, and restricted movement of articulators, among other problems, which causes severe decrease in intelligibility of the speech for which it is hard to communicate with dysarthic speakers. The aim of this thesis is to raise automatic dysarthic speech recognition rate through producing intelligibilityenhanced speech. Procedure used in which formants and energies are estimated from dysarthic speech and modified to approximately desired normal targets through modification algorithm which will have two paths. One considered with modifying the formant information through using voice conversion algorithm based on LPC coefficients replacement. The other through modification of energy trajectories of dysarthic speech as well as Fo. The modified parameters are prepared to get the final re-synthesized speech through a mean that can concatenate the modified formants speech chunk and modified energy and pitch are taken to formant synthesizer to get final transformed speech, tested through perceptual tests for ensuring quality and intelligibility, which confirmed an observed change in positively comprehending the resulting speech from the modification algorithm as well as speech recognition test using HTK word recognizer. The obtained results indicate that the applied algorithm enhances intelligibility and raises the recognition rate from 28% to 71.4% also quality of re-synthesized speech increased individuals which indicates |