Cross-Lingual Speech Depression Detection via Contrastive Alignment

Anisha Pattanayak · hf · 2026-07-08

This paper proposes a supervised contrastive alignment framework that maps speech embeddings from WavLM in English and Mandarin into a shared clinical space for cross-lingual depression detection. This approach addresses the challenge of cross-lingual generalization and reveals performance biases caused by speaker identity leakage.

Original post →

More from Research

Research channel →