Robust impaired speech segmentation using neural network mixture model

This paper presents a signal processing technique for segmenting short speech utterances into unvoiced and voiced sections and identifying points where the spectrum becomes steady. The segmentation process is part of a system for deriving musculoskeletal articulation data from disordered utterances, in order to provide training feedback for people with speech articulation problem. The approach implement a novel and innovative segmentation scheme using artificial neural network mixture model (ANNMM) for identification and capturing of the various sections of the disordered (impaired) speech signals. This paper also identify some salient features that distinguish normal speech from impaired speech of the same utterances. This research aim at developing artificial speech therapist capable of providing reliable text and audiovisual feed back progress report to the patient.