Classification of voice pathology using different features and Bi-LSTM

Voice pathology refers to any abnormality or disease affecting the human voice, resulting in voice quality deterioration. The identification and early detection of voice pathology is crucial in treating the patients effectively. In this work, the proposed system uses three feature extraction techniques, namely All-Pole Group Delay Function (APGDF), Mel-frequency cepstral coefficients (MFCC), and constant-Q cepstral coefficients (CQCC), and Bidirectional long short-term memory (BI-LSTM) is used to classify different types of pathological voices. In this work, we have adopted Saarbruecken Voice Database (SVD) database to evaluate the proposed system. The proposed system achieved an accuracy of 92.7% in classifying these pathological voice types using the CQCC feature extraction method and the BI-LSTM classification model. This system has potential applications in diagnosing and treating voice disorders.