{"ID":6496551,"CreatedAt":"2026-07-13T00:13:58.98745467Z","UpdatedAt":"2026-07-13T01:36:32.105453469Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2607.02904","arxiv_id":"2607.02904","title":"Speaker-Aware Temporal Aggregation Strategies on Segment Representations for Depression Detection in Dyadic Interaction: A Benchmark Study","abstract":"Speech-based depression detection compresses features from short audio segments into one speaker-level decision, a step called temporal aggregation rarely studied on its own. Most benchmarks fix a single self-supervised encoder and a single hand-picked layer, so a reported gain may reflect the pipeline rather than the aggregation method itself. We introduce DEPOOL, a controlled benchmark that compares six aggregation architectures with six frozen speech backbones on an English and a Mandarin depression corpus, where each configuration learns which backbone layers matter rather than fixing one by hand. Across the resulting 72-configuration grid, a third of configurations collapse into predicting a single class for every speaker, a failure tied to the backbone as much as to the method, and the architecture that is most stable in a single-seed run becomes unreliable when training repeats across seeds. Robustness to backbone and seed, rather than average accuracy across a single pipeline, should be a first-class benchmarking criterion for temporal aggregation in clinical speech.","short_abstract":"Speech-based depression detection compresses features from short audio segments into one speaker-level decision, a step called temporal aggregation rarely studied on its own. Most benchmarks fix a single self-supervised encoder and a single hand-picked layer, so a reported gain may reflect the pipeline rather than the...","url_abs":"https://arxiv.org/abs/2607.02904","url_pdf":"https://arxiv.org/pdf/2607.02904v1","authors":"[\"Anisha Pattanayak\",\"Huang-Cheng Chou\",\"Shrikanth Narayanan\",\"Sudarsana Reddy Kadiri\"]","published":"2026-07-03T03:02:48Z","proceeding":"eess.AS","tasks":"[\"eess.AS\"]","methods":"[]","has_code":false}