After an 18-month EU tender, the European Parliament deployed an AI system (RWS/CEDAT85/Bertin IT) that live-transcribes and translates plenary debates in all 24 official languages, aiding deaf and hard-of-hearing members — though no usage or accuracy stats are public.
24 languages
Official EU languages covered by live transcription/translation
18 months
Duration of the competitive tender evaluation
Details
Maturity
Established
Promoter
European Parliament (contracted to RWS / CEDAT85 / Bertin IT consortium)
Period
2020-2021 (ongoing)
Region (NUTS)
BE10
Keywords
parliamentary transparency, accessibility, machine translation, multilingual communication, public procurement
Context
The European Parliament conducts its plenary debates in 24 official EU languages, historically relying on human interpretation and same-day transcript production.
Objectives
Speed up multilingual access to plenary debate and improve accessibility for deaf and hard-of-hearing members and citizens.
Activities
Parliament ran an 18-month competitive tender for an AI-based live transcription and translation system, testing prototypes from three rival teams — Microsoft, Translated, and a consortium of RWS, CEDAT85 and Bertin IT. In November 2021 Parliament confirmed the RWS/CEDAT85/Bertin IT consortium's Language Weaver-based platform as the winner, having "ranked first throughout" the evaluation.
Results
The deployed system automatically transcribes and translates spoken floor debate in real time across all 24 languages, displaying on-screen subtitles for members — including those with hearing impairments — while accelerating release of official multilingual transcripts to the public.
Conclusions
Neither Parliament nor the winning consortium has published usage statistics, translation-accuracy figures, or citizen-reach numbers for the live service; the public record documents the tender process, technical design and selection outcome, not a post-deployment performance evaluation.
Implementation
Indicative cost
High (€500k–€5M)
Time to results
Medium (1–3 years)
Staffing & skills
European Parliament procurement and technical evaluation team, contracted vendor consortium: RWS, CEDAT85, Bertin IT
Conditions for success
a rigorous multi-vendor competitive evaluation with prototypes tested head-to-head
existing machine-translation infrastructure (Language Weaver) to build on
institutional scale that justifies the investment
Common failure modes
no public accuracy or usage reporting limits post-deployment accountability
solution scaled for one very large multilingual institution, not shown to transfer to smaller bodies
Where it fits
Governance type
EU institution, formal public-procurement process
Scale
EU-wide — all 24 official languages
Income level
high-income
Do you run this practice?
Claim it —
verified implementers get a public contact pathway and can propose corrections.
Data sources
Where this practice's information was retrieved from, and when.
India's MeitY-run BHASHINI mission gives citizens AI speech/text translation across 22 languages, embedded in e-Shram, Defence Production and Parliament portals …