Broadcaster
A broadcaster running catch-up or VOD from linear feeds that already carry compliance-approved subtitles
AI & automationExploringBuilt by Professional ServicesLast updated 29 July 2026
Teletext ExtractorSubtitle and data services recovered from broadcast feeds
Subtitles that already travel inside a broadcast feed get thrown away at the point of ingest, then paid for a second time when the same content reaches a digital platform. This POC investigates extracting those existing subtitle and data services straight from the transport stream so they can be reused instead of recreated.
The problem this solves
Teletext and DVB subtitle data are carried alongside the video in most linear feeds, and they have already been produced, checked and, where required, compliance-approved. When that feed is ingested for catch-up or VOD, the ancillary data is usually discarded because the digital pipeline expects a different subtitle format.
The consequence is that broadcasters pay twice: once for the subtitles that went out on air, and again for machine transcription or a subtitling house to reproduce them for the digital platform. The second version is often worse than the one that was thrown away.
Who this is for
This is relevant if you are:
Telco
A telco or platform operator ingesting third-party linear channels and needing subtitles on the digital tier without a separate subtitling contract
The approach we are investigating
- 1Identify which subtitle and data services a given transport stream actually carries, per channel and per language.
- 2Extract the ancillary data without re-encoding the video, so ingest performance and picture quality are untouched.
- 3Convert the extracted services into the subtitle formats a digital platform expects, preserving timing rather than re-deriving it.
- 4Establish where the recovered subtitles are good enough to publish as they are, and where a human check is still required.
What makes it different
The usual answer to missing digital subtitles is to generate new ones with speech recognition. That treats an already-solved problem as an unsolved one, and it discards the editorial and compliance work that went into the broadcast version. This POC starts from the assumption that the best subtitle available is the one that already aired, and that the job is recovery rather than generation.
Current status & next steps
ExploringReviewed 29 July 2026
This is an early concept and nothing is deployed. There is no demo on this page because there is nothing yet worth putting in front of a visitor - that is exactly what the Exploring status means here, and we would rather say so than show a placeholder.
What we are establishing first is which feed variants and subtitle service types are worth supporting, because that determines whether the effort pays back.
Does this problem exist in your workflow?
If you are paying twice for subtitles you already produced, tell us how your feeds are structured. That is the input we need to decide how far to take this.