An investigation of vocal tract characteristics for acoustic discrimination of pathological voices

Abstract

This paper investigates the effectiveness of measures related to vocal tract characteristics in classifying normal and pathological speech. Unlike conventional approaches that mainly focus on features related to the vocal source, vocal tract characteristics are examined to determine if interaction effects between vocal folds and the vocal tract can be used to detect pathological speech. Especially, this paper examines features related to formant frequencies to see if vocal tract characteristics are affected by the nature of the vocal fold-related pathology. To test this hypothesis, stationary fragments of vowel /aa/ produced by 223 normal subjects, 472 vocal fold polyp subjects, and 195 unilateral vocal cord paralysis subjects are analyzed. Based on the acoustic-articulatory relationships, phonation for pathological subjects is found to be associated with measures correlated with a raised tongue body or an advanced tongue root. Vocal tract-related features are also found to be statistically significant from the Kruskal-Wallis test in distinguishing normal and pathological speech. Classification results demonstrate that combining the formant measurements with vocal fold-related features results in improved performance in differentiating vocal pathologies including vocal polyps and unilateral vocal cord paralysis, which suggests that measures related to vocal tract characteristics may provide additional information in diagnosing vocal disorders.

Figure 1

Spectrogram of example utterances from (a) vocal fold polyp, (b) unilateral VCP, and (c) normal female. Darker regions correspond to higher energies.

Figure 2

(a) Average linear prediction (LP) spectrum and (b) spectral difference of normal and pathological female subjects.

Figure 3

(a) Average linear prediction (LP) spectrum and (b) spectral difference of normal and pathological male subjects.

Figure 4

Distribution of average formant frequency for each normal versus polyp (left) and normal versus unilateral VCP (right) subjects for females ((a) and (b)) and males ((c) and (d)). Two discriminatory formants from among F1, F2, and F3 are selected for each plot. Regions for distributions are marked with ellipses. For the case of female vocal fold polyp data, F3 mean was used in plotting the figure although its P value is larger than 0.05, because only F1 mean was found to be a significant feature from among three formant means.

Figure 5

DET curve using static and dynamic features for discriminating between normal and vocal fold polyp subjects, for (a) females and (b) males.

Figure 6

DET curve using static and dynamic features for discriminating between normal and unilateral VCP subjects, for (a) females and (b) males.

Figure 7

DET curve using formants and vocal fold-related features for discriminating between normal and vocal fold polyp subjects, for (a) females and (b) males.

Figure 8

DET curve using formants and vocal fold-related features for distinguishing between normal and unilateral VCP subjects, for (a) females and (b) males.