Sample work · Video & Animation
Live subtitles: spoken German appears as text during the broadcast
Continuous audio is recognized in short segments and displayed as subtitles, with the actual recognition latency. Technical terms come from a glossary list, and a word at the end of a segment is only shown when the next segment confirms it. Everything runs locally, without a cloud service.
Sample work (demo), not a client job. Built as a trial task with invented example data; the check figures below come from the automatic test run and are not polished.
What was built
- Recognition in short, overlapping segments so no word is lost at the boundary
- Glossary list for names and technical terms
- Output as an overlay for OBS and as a subtitle file for the recording
Subtitles with a delay of a few seconds, not a verbatim stenogram. Languages German, English and Polish.
How it was checked
Automatic test run on 2026-10-31, taken from the log (test data invented, figures measured):
- 6 windows of 4 s each over 16.1 s speech: processing time average 1.49 s, max 1.59 s
- Display delay max 5.6 s
- 100 % of words recognized
- Overlay updated 6x
- local, 0 EUR
Limits
What this sample does not cover, so nobody expects more:
- Delay typically a few seconds (4 s window + processing time), not word-for-word transcription like a stenographer
- Languages German, English, Polish; audio from customer PC (microphone/system sound) or stream source; output as overlay (OBS/browser) and WebVTT
- runs locally without cloud costs; setup on the customer's computer only with their access
- no live translation into foreign languages except de/en/pl
- technical terms/names provided in advance as a list improve recognition (glossary)
Something like this for your business?
Describe briefly what you need. You get a fixed price before I start; nothing happens without your yes.