NORMA eResearch @NCI Library

Detecting Depression from Speech with Residual Learning

Jeremiah, Donovan Michael (2020) Detecting Depression from Speech with Residual Learning. Masters thesis, Dublin, National College of Ireland.

[thumbnail of Master of Science]
PDF (Master of Science)
Download (1MB) | Preview
[thumbnail of Configuration manual]
PDF (Configuration manual)
Download (1MB) | Preview


According to the World Health Organization (WHO)1, more than 25% of the European Union’s (EU) population suffer from various levels of depression and anxiety, which if left untreated, could lead to serious health disorders such as Major Depressive Disorder (MDD) or otherwise called clinical depression. Health conditions like depression and anxiety cost the EU over e170 billion every year. This study investigates the effectiveness of residual networks in depression detection. It proposes the use of ResNet-18 to predict if an individual is depressed or not, and compares its performance to a Base CNN and AlexNet. The models are trained on the log-scaled spectrograms of the participant audio recordings from the DAICWOZ dataset. Preprocessing steps such as random undersampling and k-fold cross-validation contribute significantly to the performance of the models. The ResNet-18 model provides a substantially high F1-score of 0.83 which is 7.2% higher than the next best state-of the-art model. This research demonstrates the effectiveness of residual networks in depression detection and, hence, advocates its viable use in listening and depression helpline services. One of the limitations of the model is that it shows signs of overfitting. Future work could potentially investigate the use of General Adversarial Networks (GAN) for data augmentation techniques.

Item Type: Thesis (Masters)
Subjects: Q Science > QA Mathematics > Electronic computers. Computer science
T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science
Q Science > QA Mathematics > Computer software
T Technology > T Technology (General) > Information Technology > Computer software
Divisions: School of Computing > Master of Science in Data Analytics
Depositing User: Dan English
Date Deposited: 20 Jan 2021 15:52
Last Modified: 20 Jan 2021 15:52

Actions (login required)

View Item View Item