I work on multimodal foundation models, audio representation learning, and interpretability.

My research focuses on bridging the gap between audio and large language models, from discrete audio representations to benchmarks and evaluation. I'm finishing my PhD at Mila and Concordia University with Mirco Ravanelli and Cem Subakan. I recently completed an internship at Epic Games, working on expressive generative models. I'm a core contributor to SpeechBrain and founded the Conversational AI Reading Group.

Conversational AI Reading GroupBack on October 8! Weekly talks on speech and conversational AI, Thursdays at 11:00 ET, open to all.

See talks →

• 2026📢 Our workshop on Explainability, Robustness, and Controllability of Language Modeling for Speech and Audio was accepted at ICASSP! 🎉 Working on audio language models or conversational AI? Submit by November 11, 2026.
• Oct 2026🎙️ The Conversational AI Reading Group is back on October 8. Join us Thursdays at 11:00 ET.
• 2026Two papers accepted at Interspeech 2026: Investigating Faithfulness in Large Audio Language Models (oral) and Enhancing Audio Reasoning via Semantic Summary Prediction.
• 2026ALAS accepted at MLSP 2026.
• 2026DASB accepted at TMLR.

1. 2026

   Investigating Faithfulness in Large Audio Language Models

   Pooneh Mousavi, Lovenya Jain, Mirco Ravanelli, Cem Subakan

   Interspeech 2026 · Oral Paper Website
2. 2026

   Enhancing Audio Reasoning via Semantic Summary Prediction

   Francesco Bonzi, Pooneh Mousavi, Cem Subakan, Mirco Ravanelli

   Interspeech 2026 Paper Website
3. 2026

   DASB: Discrete Audio and Speech Benchmark

   Pooneh Mousavi, Jarod Duret, Darius Petermann, … Mirco Ravanelli

   TMLR 2026 Paper Website Code
4. 2026

   Listen First, Then Answer: Timestamp-Grounded Speech Reasoning

   Jihoon Jeong, Pooneh Mousavi, Mirco Ravanelli, Cem Subakan

   Preprint, under review Paper
5. 2026

   ALAS: An Automatic Latent Alignment Score for Audio Language Models

   Pooneh Mousavi, Yingzhi Wang, Mirco Ravanelli, Cem Subakan

   MLSP 2026 · also ICML 2025 ML4Audio Workshop Paper Code
6. 2025

   Discrete Audio Tokens: More Than a Survey!

   Pooneh Mousavi^†, Gallil Maimon^*, Adel Moumen^*, … Mirco Ravanelli

   TMLR 2025 Paper Website
7. 2025

   LiSTEN: Learning Soft Token Embeddings for Neural Audio LLMs

   Pooneh Mousavi^*, Shubham Gupta^*, Cem Subakan, Mirco Ravanelli

   Interspeech 2025 Paper
8. 2025

   What Are They Doing? Joint Audio-Speech Co-Reasoning

   Yingzhi Wang, Pooneh Mousavi, Artem Ploujnikov, Mirco Ravanelli

   ICASSP 2025 Paper
9. 2024

   How Should We Extract Discrete Audio Tokens from Self-Supervised Models?

   Pooneh Mousavi, Jarod Duret, Salah Zaiem, … Mirco Ravanelli

   Interspeech 2024 · Oral Paper Code

^* equal contribution · ^† lead author