autoclip: an AI-powered automation tool for extracting highlights and generating collections from YouTube and Bilibili videos

What it solves

AutoClip automates the tedious process of creating short-form content from long videos. It eliminates the need for manual scrubbing through footage to find highlights, manually cutting clips, and organizing them into collections for platforms like YouTube and Bilibili.

How it works

The system uses a combination of video downloading tools and Large Language Models (LLMs) to process content:

1. Ingestion: Downloads videos and subtitles from YouTube or Bilibili, or accepts local uploads.
2. AI Analysis: Uses the Qwen (Tongyi Qianwen) LLM to analyze subtitles and content to extract a video outline, identify key topic time-stamps, and score the "excitement" or quality of specific segments.
3. Processing: Automatically cuts the video into clips based on the AI-identified timestamps and generates catchy titles for each segment.
4. Management: Provides a Web interface to review clips, manually adjust sorting, and group them into AI-recommended collections.

Who it’s for

Content creators, social media managers, and video editors who want to quickly turn long-form videos into viral short-form clips or curated highlight reels.

Highlights

• Multi-platform support: Integrated downloading for YouTube and Bilibili.
• AI-driven clipping: Automatic identification of highlights and scoring of segments using LLMs.
• Asynchronous pipeline: Uses Celery and Redis to handle heavy video processing tasks in the background with real-time progress updates via WebSockets.
• Full-stack management: Includes a React-based dashboard for project management, clip editing, and collection organization.