Ericsson claims world-first latency reduction in live captioning for TV
Ericsson has unveiled its latest innovation in live captioning, developed in partnership with BBC Research & Development (BBC R&D) – a new approach that significantly reduces latency in live-captioning. Until now, live captions for television are typically displayed on screen later than the audio that they represent. The latency is caused by the necessary workflow to create live captions. This latency can be frustrating for audiences and has been subject to regulatory attention both in the UK and around the world.
Ericsson has partnered with BBC R&D to devise the new approach which minimises the delay between live captions and the audio they represent by utilizing the time taken to compress the audio and video streams for transmission and distribution. As captions take less time to encode, a compensating delay is used to ensure pre-prepared, accurately authored captions are synchronised with the audio. During programmes with live captions, this compensating delay can be decreased, which significantly reduces the apparent delay of the live captions.
Ericsson and BBC R&D have completed a successful proof of concept and the new approach will begin a phased roll out across the BBC’s portfolio of channels from summer 2016. Ericsson will also offer this service to clients globally to improve the viewing experience of audiences around the world.
Thorsten Sauer, Head of Broadcast and Media Services, Ericsson, says: “The inherent latency of live caption delivery is a challenge for broadcasters the world over. Our aim has been to improve this experience by leveraging our significant expertise in both language services as well as compression and our overall research in the media space. Together with the BBC we believe we have made some very significant advancements in the delivery and quality of live captioning and set a new bar for the future of the industry.”
Ericsson’s real-time platform supports live captioning in various languages for broadcasters around the world by using speech-to-text, newsroom integration and rapid offline content preparation. The enterprise-level, schedule-driven software platform, which was developed in-house, uses the best possible speech recognition and stenography technologies to power Ericsson’s live captioning services.
It allows multiple captioners to prepare and deliver real-time services for clients while maximizing re-use of the caption data after it has been broadcast — for example, by helping to power content discovery and enrich archive search. The platform is currently being used to deliver live captioning services for major broadcast clients around the world.