key points
Numendo supported the EBU in the design and development of EuroVOX, a real-time translation and transcription platform for European television channels. A project combining innovation, AI, and performance to serve large-scale multilingual broadcasting of audiovisual content.
1
senior full-stack lead developer deployed in Agile mode for design and technical development
1
EuroVOX multilingual platform (transcription, real-time translation
context
The European Broadcasting Union (EBU), behind the Eurovision Song Contest, is the world’s largest association of public service media broadcasters. As part of its digital transformation, it sought to facilitate the accessibility of audiovisual content across all European languages. The goal: reduce costs and complexity in transcription and translation processes while ensuring broadcast-grade quality suitable for live international events.
To address this strategic challenge, the EBU initiated the development of EuroVOX, an open multilingual transcription and translation platform built for major European TV channels.
To address this strategic challenge, the EBU initiated the development of EuroVOX, an open multilingual transcription and translation platform built for major European TV channels.
intervention : ENGINEERING A MULTILINGUAL TRANSCRIPTION AND TRANSLATION PLATFORM
Numendo provided a senior full-stack lead developer to support the EBU in the technical design and development of EuroVOX tools.
The scope of work covered both software architecture and the implementation of key platform features:
• Central API Design: Building a central API handling audio-to-text transcription, translation into over 140 languages, and real-time or batch output delivery.
• Web Interface Development: Creating a web interface allowing video uploads, transcript editing and validation, and automated file exports (text, audio, dubbing).
• Advanced AI Features: Integrating advanced capabilities such as multilingual revoicing, speaker identification, voice cloning, and live translation for conferences.
• Scoping & POC Execution: Participating in functional scoping and the Proof of Concept (POC) phase for new modules, with a constant focus on performance, reliability, and legal compliance of AI processing.
The project was delivered in Agile mode with an international team, following a “no-estimate” approach that fostered productivity and team autonomy.
The scope of work covered both software architecture and the implementation of key platform features:
• Central API Design: Building a central API handling audio-to-text transcription, translation into over 140 languages, and real-time or batch output delivery.
• Web Interface Development: Creating a web interface allowing video uploads, transcript editing and validation, and automated file exports (text, audio, dubbing).
• Advanced AI Features: Integrating advanced capabilities such as multilingual revoicing, speaker identification, voice cloning, and live translation for conferences.
• Scoping & POC Execution: Participating in functional scoping and the Proof of Concept (POC) phase for new modules, with a constant focus on performance, reliability, and legal compliance of AI processing.
The project was delivered in Agile mode with an international team, following a “no-estimate” approach that fostered productivity and team autonomy.
results
Thanks to Numendo’s expertise, the EuroVOX platform successfully transitioned out of its POC phase to be deployed in real-world conditions during major EBU events. The system is now capable of real-time content transcription and translation with precision and fluidity that meet broadcasting industry standards.
The platform is currently being rolled out across EBU member channels, serving as a strategic driver for accessibility and innovation—particularly through its revoicing and voice cloning modules, paving the way for new forms of multilingual production.
The platform is currently being rolled out across EBU member channels, serving as a strategic driver for accessibility and innovation—particularly through its revoicing and voice cloning modules, paving the way for new forms of multilingual production.
