Digital signal processing in the differential diagnosis of benign larynx diseases <b>[Abstract in English]</b>

Abstract

Aims: To present a new Digital Signal Processing procedure to evaluate voice disorders, specifically those related to the benign larynx diseases.
Methods: Recorded acoustic signals of normal and abnormal voices of outpatients subjected to videolaryngoscopy had been digitalized for the study of the alterations in vocal acoustics. Window and cepstral analysis were applied on these signals.
Results: Through the observation of the different alterations in the voices, it was possible to develop a set of parameters of analysis that allowed a coherent diagnosis of larynx alterations, based in the cepstral findings. Cases randomly selected, who had not participated of the study, had been identified only through the information of the set of parameters, with 80% of rightness, significantly validating the method.
Conclusions: This computing method is very efficient and non-invasive, successfully identifying vocal alterations by cepstral analysis.
KEY WORDS: LARYNX DISEASES; CEPSTRUM; CEPSTRAL ANALYSIS.