aiqbench analysis
This submission introduces a 1B-parameter offline model capable of simultaneous speech translation across 50 languages, demonstrating high quality and low computational overhead in simulated environments. By leveraging the Canary architecture and AlignAtt policy, the system targets practical deployment in resource-constrained settings where latency and compute efficiency are critical. The work represents a significant step toward scalable, multilingual translation tools that function without continuous internet connectivity.
A Pocket Offline Model for Simultaneous Speech Translation as CUNI Submission to IWSLT 2026
arXiv:2606.03948v1 Announce Type: new Abstract: We implement simultaneous translation capability with the offline direct speech-to-text translation model Canary, using the state-of-the-art policy AlignAtt, and submit it to IWSLT 2026 Simultaneous Speech Translation Shared task for Czech to English and English to German and Italian. The strengths of our system are: (1) high translation quality, outperforming similarly sized baselines both in low- and high-latency regimes in computationally unaware simulations; (2) low computational requirements, as the model has only 1B parameters; (3) multilinguality -- support of 25 source and 25 target languages.
Abstract:We implement simultaneous translation capability with the offline direct speech-to-text translation model Canary, using the state-of-the-art policy AlignAtt, and submit it to IWSLT 2026 Simultaneous Speech Translation Shared task for Czech to English and English to German and Italian. The strengths of our system are: (1) high translation quality, outperforming similarly sized baselines both in low- and high-latency regimes in computationally unaware simulations; (2) low computational requirements, as the model has only 1B parameters; (3) multilinguality -- support of 25 source and 25 target languages.
Computation and Language (cs.CL)
arXiv:2606.03948 [cs.CL]
(or arXiv:2606.03948v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2606.03948 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 180 words
Related coverage
SimPOTrans: Self-Adaptation for Offline-to-Simultaneous Speech Translation via Reinforcement Learning
arXiv · Oct 27, 2026
NaturalFlow: Reducing Disruptive Pauses for Natural Speech Flow in Simultaneous Speech-to-Speech Translation
arXiv · Jun 12, 2026
RASST: Retrieval-Augmented Simultaneous Speech Translation
arXiv · Jun 17, 2026
Simultaneous Speech-to-Speech Translation Without Aligned Data
arXiv · Jul 23, 2026