Comparative Evaluation of Several Pitch Process Models in the Detection of Vocal Tremor

Three models describing the speech excitation wave (denoted as pitch process) are suggested in order to detect vocal tremor in pathological speech. These models are compared, and the advantages and disadvantages of each of them are discussed. Comparison results for synthesized speech are presented, as well as spectral analysis results for real data of Parkinsonian speech, from which it turns out that the pitch process may serve as a powerful tool for detecting such tremor.