KnowClip-ai/knowclip

What it solves

KnowClip automates the process of turning long-form videos (such as courses, interviews, podcasts, or live streams) into short, engaging clips for social media. It eliminates the manual effort of scrubbing through hours of footage to find highlight moments and manually cutting them out.

How it works

The tool follows a four-step pipeline:

1. Local ASR: It uses the Paraformer-large model to perform on-device automatic speech recognition, generating transcripts with word-level timestamps.
2. LLM Analysis: The transcript is sent to an LLM (via an OpenAI-compatible API key provided by the user) to identify the most compelling segments, including titles, descriptions, and hashtags.
3. Preview: Users can review the suggested highlight segments as cards.
4. Batch Export: Selected clips are exported using ffmpeg -c copy (stream copying), which allows for near-instant export without the need for re-encoding.

Who it’s for

Content creators, educators, and podcasters who want to quickly extract highlights from long videos while maintaining privacy and keeping costs low.

Highlights

• Privacy-First: Video footage and ASR processing stay entirely on the local machine.
• Low Cost: No subscription fees; users only pay for their own LLM API usage.
• High Speed: Uses stream copying for millisecond-level clip exports.
• Flexible LLM Integration: Supports any OpenAI-compatible API, including DeepSeek, Kimi, GLM, and OpenAI.