AI use case
Turn a long episode video into six vertical captioned clips with local Whisper
A creator built a pipeline that transcribes an episode with Whisper, trims silences, cuts clips on word timestamps and burns in word-by-word captions, producing about six 9:16 clips per run. An optional step sends only the transcript, roughly 5 to 8 thousand tokens for 20 minutes, to a language model to pick clips, and a larger Whisper model gives better German captions.
- Whisper
- Marketing
- Automation
- Art
- Free tool
Source: DEV ·