And here’s the twist: No multilingual SQA data was provided in the constrained setting! Only English!

So… how did we do it? 👀

🗓️ Come find out:
Thursday, July 31st, 2PM at #IWSLT2025 #ACL2025

📄 Or read the full system report here:
arxiv.org/pdf/2506.01808

Efficient Speech Translation through Model Compression and Knowledge Distillation 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...

Deployment of large audio-language models is challenging. In our #IWSLT2025 research, we compressed Qwen2-Audio-7B to be 50% smaller, while retaining #speech translation quality. We used a mix of ✂️ iterative layer pruning, 💿quantization, and 📚knowledge distillation. #NLProc

arxiv.org/abs/2505.20237

🆕 For this year, Arabic joins the challenge!

✨ For the first time, we ask participants to generate subtitles for Arabic, covering the 5th most spoken language and one of the six UN 🌐 official languages.
Let’s break language barriers! 🌍
#Subtitling #IWSLT2025

Subtitling track Home of the IWSLT conference and SIGSLT.

Our last task: Subtitling!!

🎯 Goal: The IWSLT 2025 Subtitling Track challenges participants to generate accurate Arabic & German subtitles for English audiovisual recordings, bridging language gaps in media! 🌍📺
#AI #SpeechTech #Subtitling #IWSLT2025

🔗: iwslt.org/2025/subtitl...

Today's task is an IWSLT mainstay: Offline ST!

🎯 Goal – Provide a stable & shared evaluation framework to track advances in Spoken Language Translation from English into multiple languages & domains. 🌍🎙️

#AI #SpeechTech #SpeechTranslation #IWSLT2025