Amidst falling temperatures across Europe, one part of Belgium promises a warm welcome courtesy of EBU member VRT. Every year, in the week before the winter holidays, VRT hosts “De Warmste Week" (The Warmest Week) – a 7-day live radio broadcast from ‘t Zand Square in Bruges, raising money and awareness for a range of different charities and causes. This year’s coverage will be more accessible than ever, and to a larger potential audience, thanks to live captions provided by the EBU EuroVOX project.
Running 18-24 December, the VRT campaign seeks to mobilise all of Belgium around the theme of “growing up without worry". It aims to highlight issues faced by children and young adults and promotes the vision of a "care-free youth" where every child should feel safe and secure when growing up.
The audiences of the VRT media coverage will be able to see realtime captions of what is being said on-air throughout the entire event, on the digital platform VRT MAX as well as some smart TV sets.
After successfully running the system at EBU events and assemblies, where the same technology was applied (coupled with live translation to deliver captions in up to 40 additional languages) for audiences on site, the opportunity to show how this could also be applied to broadcast captioning was the next logical step.
By taking a low-latency audio stream directly from a broadcasters own infrastructure, EuroVOX is able to efficiently use a range of different vendors to perform both transcription and translation to produce captions in the right language. This is then provided back as a feed that can be integrated back into a captioning workflow, with a latency of around 2-3s.
The VRT's project lead, Klaas Baert, said: "With Warmste Week we are shooting for 100% accessibility. At the same time we want to experiment how AI can help us to achieve that. With EuroVOX, EBU brings their PSM domain knowledge in the mix to easily integrate speech-to-text in the live workflow of Warmste Week"
This latest advance is another in a set of advances that will eventually enable users of the EuroVOX studio to be able to easily translate and transcribe both file-based and live content in a hybrid workflow, expected to be available later in 2024.
Contact us for more information!