Disease Detection Using Analysis of Voice Parameters

This paper investigates the adaptation of automatic speech recognition to disease detection by analyzing the voice parameters. The analysis of the voice allows the identification of the diseases which affect the vocal apparatus and currently is carried out from an expert doctor through methods based on the auditory analysis. This paper presents a novel method to keep track of patient’s pathology: Easy to use, fast ,non invasive for the patient and affordable for the clinician. This method uses parametric method (jitter, shimmer, harmonic to noise etc...) to evaluate the pathological voice. The method for this task also relies on Mel Frequency Cepstral Coefficient (MFCC) as feature extraction and Dynamic Time Warping (DTW) as feature Matching. The aim of the study is to evaluate the voice quality in patients with mild-to-acute asthma by parametric method and non parametric method. Comparative analysis is also done between parametric and non-parametric methods. Keywords—MFCC, DTW, Dysphonia, Jitter, shimmer, HNR, Acoustic parameters.