Efficient Speech Translation through Model Compression and Knowledge Distillation

Yasmin Moslem


Abstract
Efficient deployment of large audio-language models for speech translation remains challenging due to their significant computational requirements. In this paper, we address this challenge through our system submissions to the ‘Model Compression’ track at the International Conference on Spoken Language Translation (IWSLT 2025). We experiment with a combination of approaches including iterative layer pruning based on layer importance evaluation, low-rank adaptation with 4-bit quantization (QLoRA), and knowledge distillation. In our experiments, we use Qwen2-Audio-7B-Instruct for speech translation into German and Chinese. Our pruned (student) models achieve up to a 50% reduction in both model parameters and storage footprint, while retaining 97-100% of the translation quality of the in-domain (teacher) models.
Anthology ID:
2025.iwslt-1.40
Volume:
Proceedings of the 22nd International Conference on Spoken Language Translation (IWSLT 2025)
Month:
July
Year:
2025
Address:
Vienna, Austria (in-person and online)
Editors:
Elizabeth Salesky, Marcello Federico, Antonis Anastasopoulos
Venues:
IWSLT | WS
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
379–388
Language:
URL:
https://preview.aclanthology.org/acl25-workshop-ingestion/2025.iwslt-1.40/
DOI:
Bibkey:
Cite (ACL):
Yasmin Moslem. 2025. Efficient Speech Translation through Model Compression and Knowledge Distillation. In Proceedings of the 22nd International Conference on Spoken Language Translation (IWSLT 2025), pages 379–388, Vienna, Austria (in-person and online). Association for Computational Linguistics.
Cite (Informal):
Efficient Speech Translation through Model Compression and Knowledge Distillation (Moslem, IWSLT 2025)
Copy Citation:
PDF:
https://preview.aclanthology.org/acl25-workshop-ingestion/2025.iwslt-1.40.pdf