Spectral analysis of pathological acoustic speech waveforms .

摘要

Biomedical engineering is the application of engineering principles and techniques to the medical field. The design and problem solving skills of engineering are combined with medical and biological science, which improves medical disorder diagnosis and treatment. The purpose of this study is to develop an automated procedure for detecting excessive jitter in speech signals, which is useful for differentiating normal from pathologic speech. The fundamental motivation for this research is that tools are needed by speech pathologists and laryngologists for use in the early detection and treatment of laryngeal disorders. Acoustical analysis of speech was performed to analyze various features of a speech signal. Earlier research established a relation between pitch period jitter and harmonic bandwidth. This concept was used for detecting laryngeal disorders in speech since pathologic speech has been found to have larger amounts of jitter than normal speech.;Our study was performed using vowel samples from the voice disorder database recorded at the Massachusetts Eye and Ear Infirmary (MEEI) in1994. The KAYPENTAX company markets this database. Software development was conducted using MATLAB, a user-friendly programming language which has been applied widely for signal processing. An algorithm was developed to compute harmonic bandwidths for various speech samples of sustained vowel sounds. Open and closed tests were conducted on 23 samples of pathologic and normal speech samples each. Classification results showed 69.56% probability of correct detection of pathologic speech samples during an open test.