Formant analysis in dysphonic patients and automatic Arabic digit speech recognition

Abstract

Background and objectiveThere has been a growing interest in objective assessment of speech in dysphonic patients for the classification of the type and severity of voice pathologies using automatic speech recognition (ASR). The aim of this work was to study the accuracy of the conventional ASR system (with Mel frequency cepstral coefficients (MFCCs) based front end and hidden Markov model (HMM) based back end) in recognizing the speech characteristics of people with pathological voice.Materials and methodsThe speech samples of 62 dysphonic patients with six different types of voice disorders and 50 normal subjects were analyzed. The Arabic spoken digits were taken as an input. The distribution of the first four formants of the vowel /a/ was extracted to examine deviation of the formants from normal.ResultsThere was 100% recognition accuracy obtained for Arabic digits spoken by normal speakers. However, there was a significant loss of accuracy in the classifications while spoken by voice disordered subjects. Moreover, no significant improvement in ASR performance was achieved after assessing a subset of the individuals with disordered voices who underwent treatment.ConclusionThe results of this study revealed that the current ASR technique is not a reliable tool in recognizing the speech of dysphonic patients.

Acurácia das medidas acústicas tradicionais e formânticas na avaliação da qualidade vocal

“…It may be inferred from these findings that compared to individuals without voice quality deviation, patients with a mild to moderate degree of deviation may implement supraglottic adjustments to compensate for dysfunctional glottic conditions with the presence of increased silent airflow. These findings are consistent with other studies (8)(9)(10)(13)(14)(15) showing that dysphonic patients tend to make adjustments in the vocal tract to compensate for their voice problem.…”

“…Some studies (8)(9)(10)(13)(14)(15) have observed that patients with a voice disorder make adjustments not just in the glottis but also in the supraglottis. These patients have lower formant values (F1, F2, F3) than individuals without a voice disorder (10,13,15) .…”

A Novel Method for Classifying Body Mass Index on the Basis of Speech Signals for Future Clinical Applications: A Pilot Study

“…For speech feature extraction, we extracted 65 features from the collected data set. The extracted features consisted of pitch, average ratio of pitch period, correlation coefficient between F0 and intensity (CORR), absolute Jitter (Jita), and Mel frequency cepstral coefficients (MFCC), among others [18, 23, 27]. The specific content of the extracted features is described in Table 1, and sample of speech signal recording of 5 vowels and one sentence is showed in Figure 1.…”

Multidirectional Regression (MDR)-Based Features for Automatic Voice Disorder Detection

“…20 It was found that the learning of vector quantization methodology performed better than the multilayer perceptron architecture in voice pathology detection with the sustained vowel jɑj. ASR system for Arabic words with various dysphonic patients was studied by Muhammad et al 21 They reported that speech samples from sulcus vocalis patients were recognized the least (56%), whereas that from patients with vocal fold nodules were better recognized (84.5%). Wavelet decomposition and neural network-based voice pathology detection were proposed by Salhi et al 22 Energy from different subbands of wavelet was used as features.…”