23CSM1R19_Mtech_CSE_DLProject_voice.pptx

voice depression

1. Depression Detection By Sridhar Padala 23CSM1R19 Guided by U Shivani Sri Varshini

2. Introduction Depression , a common mental health problem, which is often caused by personal and professional stress and can lead to severe consequences, including suicidal thoughts.

3. Problem Statement Many cases of depression are not identified early, leading to prolonged suffering and worsened symptoms due to lack of routine screening and stigma. Depression often presents with a wide range of symptoms that vary in severity and manifestation, making it challenging to recognize and diagnose.

4. Existing Model

5. Existing Model

6. Proposed Model

7. Results Accuracy Recall Precision SPE F1 Score 73.0 66.1 53.6 75.9 59.2

8. Dataset Description The DAIC-WOZ dataset, includes transcripts of clinical interviews conducted by trained professionals with individuals diagnosed with depression.  This includes 189 sessions of interactions ranging between 7-33 minutes (average is 16 minutes).

9. References [1] L. Cao, H. Zhang, and L. Feng, “Building and using personal knowledge graph to improve suicidal ideation detection on social media,” IEEE Trans. Multimedia, vol. 24, pp. 87–102, 2022. [2] X. Xu et al., “Exploring zero-shot emotion recognition in speech using semantic-embedding prototypes,” IEEE Trans. Multimedia, vol. 24, pp. 2752–2765, 2022. [3] J. Zheng, S. Zhang, Z. Wang, X. Wang, and Z. Zeng, “Multi-channel weight-sharing autoencoder based on cascade multi-head attention for multimodal emotion recognition,” IEEE Trans. Multimedia, vol. 25, pp. 2213–2225, 2023. [4] W. C. de Melo, E. Granger, and M. B. Lopez, “MDN: A deep maximization-differentiation network for spatio -temporal depression detection,” IEEE Trans. Affect. Comput ., vol. 14, no. 1, pp. 578–590, Jan.– Mar. 2023.

10. Thank You