Elamur Research⌕
☰
home / papers / 2026 / COPECO-Speech: Multimodal Post-Editing with Speech and LLMs for Translation Teaching

COPECO-Speech: Multimodal Post-Editing with Speech and LLMs for Translation Teaching

Summary

Translator training must keep up with machine translation and LLMs, yet widely used CAT tools such as MateCat are mostly keyboard-driven, offer only basic speech features and do not support LLM-driven editing. This demo paper presents COPECO-Speech, an extension of the University of Geneva's pedagogical post-editing platform COPECO. It adds spoken interaction and LLM-assisted editing for classroom use. Instructors upload source texts with MT output from any system, assign tasks and configure prompts and settings. Students can post-edit by typing or by voice. A few fixed spoken commands handle navigation. Free-form spoken instructions, such as insertions, deletions, reformulations or full re-translation, are transcribed and sent to an LLM along with the source and MT, and the LLM returns a revised segment. The platform logs every interaction, and teachers can annotate work with shared or personalised schemes. The paper reports no user study or quantitative evaluation. Its contribution is the tool itself, which helps instructors examine and discuss how students combine human editing with AI assistance.

Key points

  • COPECO-Speech is the first integration of LLM functionality into the COPECO platform (copeco.unige.ch). It offers typing as a baseline alongside speech/LLM-based post-editing, and users can switch between modes.
  • ASR is provided by Azure Speech services and editing by GPT-4.1-mini, guided by prompts covering language detection, targeted PE or re-translation, and output consistency. Both the ASR and LLM components can be replaced.
  • Hard-coded voice commands for navigation include "Next", "Previous", "Save segment" and "Undo". Editing requests like "insert A before B" are interpreted by the LLM.
  • Logs capture keystrokes, spoken input, time spent, teacher corrections, LLM requests and outputs, and the percentage of MT edited. Reports include PE statistics and comparison with reference translations.
  • For now, only English and French are available, though other language pairs could be added. It was presented at TAITT 2026 as a demo, without an empirical evaluation.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Linguistics & languages · Machine translation

Cite

The paper itself, not this page. BibTeX from the library's record of the work.

@inproceedings{liyanapathirana2026copeco,
  title = {{COPECO-Speech: Multimodal Post-Editing with Speech and LLMs for Translation Teaching}},
  author = {Liyanapathirana, Jeevanthi and Bouillon, Pierrette and Mutal, Jonathan and Girletti, Sabrina and Volkart, Lise},
  booktitle = {Proceedings of the 1st International Workshop on Teaching AI-Based Translation and Technologies (TAITT 2026)},
  year = {2026},
  url = {https://aclanthology.org/2026.taitt-1.6/}
}

References

7 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Bowker, Lynne and Jairo Buitrago Ciro. 2019. Machine Translation and Global Research: Towards Improved Machine Translation Literacy in the Scholarly Community. Emerald Publishing Limited, Bingley. find
  2. Carl, Michael, Moritz Schaeffer, and Srinivas Bangalore. 2016. The CRITT translation process research database. In New Directions in Empirical Translation Process Research: Exploring the CRITT TPR-DB, pages 13–54. Springer, Cham. find
  3. Castaldo, Antonio, Sheila Castilho, Joss Moorkens, and Johanna Monti. 2025. Extending CREAMT: Leveraging large language models for literary translation post-editing. In Proceedings of Machine Translation Summit XIX, page 506–515. European Association for Machine Translation. find
  4. Jiao, Wenxiang, Wenxuan Wang, Jen-tse Huang, Xing Wang, and Zhaopeng Tu. 2023. Is ChatGPT a good translator? A preliminary study. arXiv preprint arXiv:2301.08745. find our card
  5. Koponen, Maarit. 2016. Is machine translation post-editing worth the effort? A survey of research into post-editing and effort. The Journal of Specialised Translation, (25):131–148. find
  6. Liyana Pathirana, Jeevanthi, Perrine Schumacher, Pierrette Bouillon, and Jonathan David Mutal. 2024. Enhanced multilingual speech-based post-editing based on user feedback with COPECO-SPEECH. In Proceedings of Translating and the Computer 46, pages 161–167, Luxembourg. Tradulex. find
  7. Mutal, Jonathan, Pierrette Bouillon, Perrine Schumacher, and Johanna Gerlach. 2020. COPECO: a collaborative post-editing corpus in pedagogical context. In Proceedings of the Workshop on Post-Editing Technology and Practice (PEMDT) at AMTA, pages 61–78. find

This page in Markdown → https://md.research.elamur.ai/free/en/papers/liyanapathirana-2026-copeco-speech-multimodal-post-editing/