Document Details

Document Type : Thesis 
Document Title :
Detecting Depression in Arabic Speech using Speech Language Recognition
الكشف عن الاكتئاب في الكلام العربي باستخدام التعرف على لغة الكلام
 
Subject : Faculty of Computing and Information Technology 
Document Language : Arabic 
Abstract : Depression is one of the most common mental illnesses. Inaccurate assessments and misdiagnosis of the illness is quite common for such mental disorders. In response to the issue of inaccurate assessment and misdiagnosis of depression, this study discusses the use of speech-language recognition to improve the detection of depression in Arabic speech. In this study, we extract speech features after collecting the dataset. Those speech features can be obtained from both linguistic (uttered words) and para-linguistic (acoustic cues) features, which we focus on. The participants are classified into two groups: clinically depressed and non-depressed groups. To do that, we start by recording speeches from interviews for the two selected groups. Then we extract para-linguistic features by using Mel- frequency cepstral coefficients (MFCC) which works well with audio data and as a result of the convolutional neural network (CNN) model, the test accuracy reached 98% which could be considered effective in detecting depression in audio speech. Moreover, Textual data was also extracted through machine learning techniques using Bags of Words (BOW) and Term Frequency-Inverse Document Frequency (TF-IDF). The algorithms utilised in predicting the presence of depression in textual data (speech) involved a Support Vector Machine, Naïve Bayes, Decision Tree, Random Forest and yielded varying levels of accuracy and precision and a majority had scores above 50%. The study outcomes show significant potential in the use of speech-language recognition in detecting depression using both audio and textual data. There is thus, the need for mental health institutions to include speech recognition techniques in detecting depression among their clients to effectively diagnose mental health issues. 
Supervisor : Dr. Salma Elhag 
Thesis Type : Master Thesis 
Publishing Year : 1444 AH
2023 AD 
Co-Supervisor : Dr.Sulhi Alfakeh 
Added Date : Tuesday, September 12, 2023 

Researchers

Researcher Name (Arabic)Researcher Name (English)Researcher TypeDr GradeEmail
زينب خليفه الشريفAlsharif, Zainab KhalifhaResearcherMaster 

Files

File NameTypeDescription
 49305.pdf pdf 

Back To Researches Page