If you are a video conferencing company dealing with multilingual users struggling to communicate across language barriers — this project developed a live automatic subtitling and translation system tested with alfaview, a German online conferencing platform. It handles dozens of source and target languages simultaneously, turning any call into a multilingual meeting without human interpreters.
Live Translation and Auto-Summaries for Multilingual Meetings and Conferences
Imagine sitting in a conference where the speaker talks in Czech, but you see live subtitles in your own language — and after the meeting, you get a neat summary instead of having to take notes. That's what ELITR built. They created software that listens to live speech, translates it on the fly into dozens of languages, and even writes automatic meeting minutes. It was tested in real settings: the Czech national audit office and a German video conferencing platform called alfaview.
What needed solving
Companies and institutions operating across European languages spend thousands on professional interpreters for conferences, meetings, and workshops — and still lose information because meeting minutes depend on manual note-taking. Online conferencing platforms lack real-time multilingual translation, forcing participants into English-only or single-language silos.
What was built
The project built a live automatic subtitling and spoken language translation system for conferences and online meetings, plus an automatic minuting tool that generates structured summaries from speech. Two demonstrators were delivered: the Automatic Minuting Demonstrator and a Publishing Platform, with 23 deliverables total.
Who needs this
Who can put this to work
If you are a government body dealing with multilingual proceedings and the cost of professional interpreters at large assemblies — this project developed automatic spoken language translation tested by the Supreme Audit Office of the Czech Republic. It also produces automatic meeting minutes from live discussions, cutting the time and cost of post-meeting documentation.
If you are a conference organizer dealing with expensive simultaneous interpretation services and delayed transcript delivery — this project developed a live subtitling system for conference presentations that works across dozens of languages. The automatic minuting demonstrator generates structured summaries directly from live speech, eliminating manual note-taking.
Quick answers
What would it cost to deploy this translation system compared to hiring human interpreters?
The project data does not include specific pricing. However, the system was designed to replace or supplement live interpreting at conferences and meetings, which typically costs hundreds of euros per hour per language pair. A software-based solution tested across dozens of languages could drastically cut per-event translation costs.
Can this scale to large conferences with hundreds of attendees and dozens of languages?
Yes — the project explicitly aimed to scale up to multilingual settings with dozens of source and/or target languages. It was tested at large assemblies like conferences as well as smaller workshops and online meetings, demonstrating flexibility across event sizes.
Who owns the intellectual property and can I license this technology?
The project was coordinated by Univerzita Karlova (Charles University, Czech Republic) under a 6-partner consortium across 4 countries. IP terms depend on the consortium agreement. Any licensing discussion would need to go through the coordinator or the relevant industrial partners.
How accurate is the automatic translation and minuting?
The project produced 23 deliverables including a dedicated Demonstrator of Automatic Minuting, indicating significant development effort. The system aimed to deliver machine translation of high-quality at the level of whole documents. Specific accuracy benchmarks are not included in the available project data.
What languages are supported?
The project was designed to handle dozens of source and/or target languages. The consortium included partners from 4 countries (Czech Republic, Germany, Italy, UK), and the core goal was removing language barriers among European citizens and organizations.
Is this ready for production use today?
The project closed in March 2022 and produced working demonstrators including the Automatic Minuting system and a Publishing Platform. It was pilot-tested by the Supreme Audit Office of the Czech Republic and alfaview, a commercial German conferencing platform, suggesting near-production readiness.
Can this integrate with our existing conferencing tools?
The system was specifically tested with alfaview, a commercial online conferencing platform, which demonstrates integration capability with real-world video conferencing software. Based on available project data, the architecture was designed for both live in-person events and online meetings.
Who built it
The ELITR consortium brings together 6 partners from 4 countries (Czech Republic, Germany, Italy, UK) with a 33% industry ratio — 2 industry players alongside 3 universities and 1 other organization. The coordinator is Charles University in Prague, a leading research institution in computational linguistics. The presence of alfaview as an industrial partner and pilot tester is a strong commercialization signal — it means the technology was built with a real product integration in mind, not just as an academic exercise. No SMEs are in the consortium, which means licensing or partnerships would likely go through the established industry partners or the university's technology transfer office.
- UNIVERZITA KARLOVACoordinator · CZ
- KARLSRUHER INSTITUT FUER TECHNOLOGIEparticipant · DE
- THE UNIVERSITY OF EDINBURGHparticipant · UK
Univerzita Karlova (Charles University), Czech Republic — reach out via their Faculty of Mathematics and Physics, Institute of Formal and Applied Linguistics
Talk to the team behind this work.
Want an introduction to the ELITR team to discuss licensing or integration? SciTransfer can arrange a direct meeting with the right people.