Overview & Result
jevmeter is an open-source media editing tool that overlays a live animated metric gauge onto video footage. It transcribes spoken dialogue, scores each sentence through Jev against a user-defined perspective (e.g., optimism, technical rigor, confidence), and renders an edited 16:9 video with dynamic visual meters.
How Jev fits in the loop
- Extracts and transcribes audio timestamps into discrete sentence intervals.
- Dispatches sentences to Jev for fast numerical scoring against custom criteria.
- Renders dynamic overlay graphics onto the source video using FFmpeg.
How to reproduce
- Clone https://github.com/ChetasLua/jevmeter and install requirements.
- Run jevmeter on an MP4 video specifying the scoring criteria: jevmeter --video talk.mp4 --metric optimism.
- Inspect the generated 16:9 output video featuring dynamic animated gauges over each spoken sentence.
Why this build matters
Demonstrates automated video content augmentation by coupling fast speech transcription with rapid Jev scoring.
Limitations
- Requires FFmpeg and speech-to-text models installed locally.
- Scoring reflects defined prompt criteria rather than objective truth.