Media & public communication
Automating recording, cutting and captioning for a livestream
Public institution (client confidential)
Problem
A public institution needed its livestream recorded, split into publishable clips, and captioned — all done by hand for every stream, which limited how much content the team could realistically publish.
Solution
I built a system that records the stream continuously, automatically saves it in 5-minute segments so any portion can be cut freely, and uses automatic speech recognition (via a Luxembourgish ASR service) to generate titles and subtitles, which are then burned into the video in the desired font — all without manual intervention.
Result
- 24/7
- automatic stream recording
- Auto
- titles & subtitles generated and burned in
- 0
- manual re-typing of captions
The team can now pull a ready-to-publish, captioned clip from any 5-minute segment of the stream without touching a transcription or subtitling tool — work that used to happen by hand for every clip now happens automatically in the background.
Technical details
- Architecture
- A recording service continuously captures the stream and writes it to storage in fixed-length segments; an editing interface lets the team select and cut arbitrary ranges, which are then passed through an ASR pipeline for transcription and subtitle burn-in.
- Technologies
- TypeScript · Node.js · FFmpeg
- Integrations & APIs
- Luxembourgish speech-to-text (ASR) service · Livestream source
- AI / decision logic
- Automatic speech recognition (Luxembourgish ASR) drives title and subtitle generation directly from the audio track.
- Deployment
- Runs continuously on managed infrastructure, recording and segmenting the stream in the background.
Does this sound like your business?
Tell me which process costs your team the most time and I will tell you honestly whether it can be removed.
Get Your Free Analysis