The eight TV2 regions that serve the Danish population have all installed SysMedia WinCAPS Live subtitling systems to provide subtitles for their regional news bulletins.
In response to government mandates on behalf of deaf and hard of hearing viewers, all TV2 regions now provide the required subtitles on their main local news bulletins from 19:30 to 20:00 each weekday evening.
As part of the new systems, a WinCAPS newsroom interface imports scripts directly from the local iNews or ENPS newsroom system. Live subtitling of unscripted content is performed using the respeaking technique with the Nuance Speech Magic speech recognition engine, as used at DR and TV2 Denmark, feeding the dedicated SpeakTITLE module in WinCAPS.
Olav Helms, Technical Project Manager with TV2 Midt-Vest, said: “We use a combination of live (re-spoken) subtitles and subtitles based on the scripts from the newsroom systems. Quite a lot of the news is prepared very close to-air, so up to 80% of the content may need to be re-spoken. If there’s time in advance then of course we prepare as much as we can.”
In some of the regions there’s an 8-10 minute feature slot and subtitles for this are prepared in advance. Helms continued: “WinCAPS handles the switch between live re-speaking and prepared subtitles very well and with minimum effort, which is important to us as we have to do it several times during a broadcast. ”
John Boulton, SysMedia’s Subtitling Products Manager, added: “The particular challenges of subtitling news and other live programmes are well known to SysMedia. Having previously provided systems to both DR and TV2/Denmark, it’s very pleasing to see WinCAPS now being adopted by the TV2 regions as well. Subtitling target levels for deaf and hard of hearing viewers are growing across Europe, affecting both national and regional broadcasters. Speech recognition technology is a vital ingredient for live subtitling, and we urge more national governments to follow the Danish lead when setting mandated subtitling obligations, and also to help finance development of suitable language models if they do not already exist.”