Durga Devi A

Bio: Durga Devi A is an academic researcher. The author has contributed to research in topics: Voice analysis & Voice activity detection. The author has an hindex of 1, co-authored 1 publications receiving 15 citations.

Pathological Voice Recognition for Vocal Fold Disease

Abstract: Pathology is the study and diagnosis of disease. Due to the nature of job, unhealthy habits and voice abuse, the people are subjected to the risk of voice problems. The diagnosis of vocal and voice disorders should be in the early stage otherwise it causes changes in the normal signal. It is well known that most of vocal fold pathologies cause changes in the acoustic voice signal. Therefore, the voice signal can be a useful tool to diagnose them. Acoustic voice analysis can be used to characterize the pathological voices. This paper presents the detection of vocal fold pathology with the aid of the speech signal recorded from the patients. We are going to recognize the disordered voice for vocal fold disease by focusing on the classification of pathological voice from healthy voice based on acoustic features. The method includes two steps. The first step is the extraction of feature vectors based on MFCC. The second is the classification of feature vectors using GMM. The extracted acoustic parameters from the voice signals are used as an input for the MFCC. The main advantage of this method is less computation time and possibility of real-time system development. This report introduces the design and implementation of the proposed system for recognizing pathological and normal voice. Also a description is given about the literature survey done and the implementation of different modules in the system. The result of the proposed system and the scope of improvements are also discussed in the report.

Voice Pathologies Classification and Detection Using EMD-DWT Analysis Based on Higher Order Statistic Features

Abstract: Background The voice is a prominent tool allowing people to communicate and to change information in their daily activities. However, any slight alteration in the voice production system may affect the voice quality. Over the last years, researchers in biomedical engineering field worked to develop a robust automatic system that may help clinicians to perform a preventive diagnosis in order to detect the voice pathologies in an early stage. Method In this context, pathological voice detection and classification method based on EMD-DWT analysis and Higher Order Statistics (HOS) features, is proposed. Also DWT coefficients features are extracted and tested. To carry out our experiments a wide subset of voice signal from normal subjects and subjects which suffer from the five most frequent pathologies in the Saarbrucken Voice Database (SVD), is selected. In The first step, we applied the Empirical Mode Decomposition (EMD) to the voice signal. Afterwards, among the obtained candidates of Intrinsic Mode Functions (IMFs), we choose the robust one based on temporal energy criterion. In the second step, the selected IMF was decomposed via the Discrete Wavelet Transform (DWT). As a result, two features vector includes six HOSs parameters, and a features vector includes six DWT features were formed from both approximation and detail coefficients. In order to classify the obtained data a support vector machine (SVM) is employed. After having trained the proposed system using the SVD database, the system was evaluated using voice signals of volunteer's subjects from the Neurological department of RABTA Hospital of Tunis. Results The proposed method gives promising results in pathological voices detection. The accuracies reached 99.26% using HOS features and 93.1% using DWT features for SVD database. In the classification, an accuracy of 100% was reached for “Funktionelle Dysphonia vs. Rekrrensparese” based on HOS features. Nevertheless, using DWT features the accuracy achieved was 90.32% for “Hyperfunktionelle Dysphonia vs. Rekurrensparse”. Furthermore, in the validation the accuracies reached were 94.82%, 91.37% for HOS and DWT features, respectively. In the classification the highest accuracies reached were for classifying “Parkinson versus Paralysis” 94.44% and 88.87% based on HOS and DWT features, respectively. Conclusion HOS features show promising results in the automatic voice pathology detection and classification compared to DWT features. Thus, it can reliably be used as noninvasive tool to assist clinical evaluation for pathological voices identification.

Vocal folds pathologies classification using Naïve Bayes Networks

Abstract: in this study the Nave Bayes Network NBN classifier is used for automatic vocal folds pathologies detection and classification. The proposed method is based on the acoustic parameters extraction such as Mel Frequency Cepstral Coefficient (MFCC), jitter, shimmer and fundamental frequency which are used as inputs to NBN classifier to discriminate between three different groups: speakers with normal voice, speakers with spasmodic dysphonia and speakers with vocal folds paralysis. For classification we used a variety of voice simples (signal of vowels production) containing simples of the three groups mentioned. Our study is developed around Saarbruecken Voice Database (SVD) it is an open German database containing deferent samples, words, sentences of normal and pathological voice. The classification rate of the developed detection system is 90%.

Phoneme independent pathological voice detection using wavelet based MFCCs, GMM-SVM hybrid classifier

Abstract: The paper proposes a new method for the phoneme independent normal and pathological voice classification. The new method proposes a wavelet sub band based hybrid classifier built by combining Gaussian Mixture Model-Universal Background Model (GMM-UBM) and Support Vector Machine (SVM). The Mel Frequency Cepstral Coefficients (MFCCs) are computed for each sub band obtained by wavelet decomposition. The MFCCs of each sub band are modelled using GMM-UBM. Finally the scores of GMM-UBMs are fused using SVM. The fusion of GMM -UBM for wavelet sub band MFCCs and SVM gives a maximum accuracy of 96.61% whereas conventional MFCCs with GMM -UBM gives 85.18%.