# Speech based Depression Severity Level Classification Using a Multi-Stage Dilated CNN-LSTM Model

arXiv: https://arxiv.org/abs/2104.04195 · HTML: https://mlsanity.com/paper/2104.04195/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Builds on 1 of its 34 references that are in the map.
- First posted on arXiv 9 Apr 2021.

## Builds on (most cited first, 1 of 1)

- [Multi-level Attention network using text, audio and video for Depression Prediction](https://mlsanity.com/paper/1909.01417/) (2019)

## Built on by (most cited first, 4 of 4)

- [Multimodal Depression Classification Using Articulatory Coordination Features And Hierarchical Attention Based Text Embeddings](https://mlsanity.com/paper/2202.06238/) (2022)
- [Speech-based Clinical Depression Screening: An Empirical Study](https://mlsanity.com/paper/2406.03510/) (2024)
- [Acoustic to Articulatory Inversion of Speech; Data Driven Approaches, Challenges, Applications, and Future Scope](https://mlsanity.com/paper/2504.13308/) (2025)
- [Tracking Articulatory Dynamics in Speech with a Fixed-Weight BiLSTM-CNN Architecture](https://mlsanity.com/paper/2504.18099/) (2025)

## Abstract (arXiv)

Speech based depression classification has gained immense popularity over the recent years. However, most of the classification studies have focused on binary classification to distinguish depressed subjects from non-depressed subjects. In this paper, we formulate the depression classification task as a severity level classification problem to provide more granularity to the classification outcomes. We use articulatory coordination features (ACFs) developed to capture the changes of neuromotor coordination that happens as a result of psychomotor slowing, a necessary feature of Major Depressive Disorder. The ACFs derived from the vocal tract variables (TVs) are used to train a dilated Convolutional Neural Network based depression classification model to obtain segment-level predictions. Then, we propose a Recurrent Neural Network based approach to obtain session-level predictions from segment-level predictions. We show that strengths of the segment-wise classifier are amplified when a session-wise classifier is trained on embeddings obtained from it. The model trained on ACFs derived from TVs show relative improvement of 27.47% in Unweighted Average Recall (UAR) at the session-level classification task, compared to the ACFs derived from Mel Frequency Cepstral Coefficients (MFCCs).