Cepstral analysis of speech signals in the process of automatic pathological voice assessment Anna

The paper describes the problem of cepstral speech analysis in the process of automated voice disorder probability estimation. The author proposes to derive two of the most diagnostically significant voice features: quality of harmonic structure and degree of subharmonic from cepstrum of speech signal. Traditionally, these attributes are estimated auricularly or by spectrum (or spectrogram) observation, hence this analysis often lacks accuracy and objectivity. The introduced parameters were calculated for the recordings from Disordered Voice Database (Kay, model 4337 version 2.7.0) which consists of 710 voice samples (657 pathological, 53 healthy) recorded in the laboratory environment and described with diagnosis and a number of additional attributes (such as age, sex, nationality). The proposed cepstral voice features were compared to similar voice parameters derived from Multidimensional Voice Program (Kay, model 5105 version 2.7.0) in respect to their diagnostic significance and presented graphically. The results show that cepstral features are more correlated with decision and better discriminate clusters of healthy and disordered voices. Additionally, both parameters are obtained by single cepstral transform and do not require to perform F0 tracking earlier as it is derived simultaneously.