The main trade-off in automated clipping is often hidden: a hosted service may simplify processing, but it also moves more of the workflow off your machine. This project takes the opposite approach—AI helps select and package moments, while cutting, framing, and rendering remain local. That makes it a practical option when you want repeatable short-form output without handing the whole video to a cloud renderer.
What Sets It Apart
- It analyzes the transcript to find complete moments, favoring coherent questions and answers over arbitrary cuts. That helps clips make sense without requiring a manual timeline edit for every candidate.
- Platform-specific rules shape the output: vertical interview or podcast layouts for short-video platforms, and landscape for Bilibili or YouTube. Each clip can include captions, a cover, metadata, and a publish kit.
- You can use creator subtitles, local speech recognition, or configured cloud transcription, and choose local or hosted language models for analysis. The choice affects privacy, setup effort, and cost.
- Desktop, CLI, and MCP share the quick-output workflow, so the same clipping process can serve individual creators or automated pipelines.
Who It’s For—and the Trade-Off
A good fit for creators repurposing interviews, podcasts, lectures, and talking-head videos who want platform-ready drafts and control over local processing. It is less suitable if you expect a fully managed cloud service, instant results without model setup, or reliable automated selection without reviewing the clips. Cloud features require your own provider credentials and may incur usage charges; local processing depends on your computer and speech-model setup.