FujiwaraChoki/supoclip

What it solves

SupoClip addresses the limitations of proprietary AI video clipping tools, such as restrictive pricing plans, platform lock-in, and mandatory watermarks. It provides an open-source alternative for creators to transform long-form videos (podcasts, streams, talks) into short, viral-ready vertical clips without these constraints.

How it works

The tool processes long videos through a pipeline that combines several AI technologies:

1. Transcription & Selection: It uses AssemblyAI for word-level transcription and an LLM (such as Gemini, GPT, Claude, or local Ollama models) to identify the most engaging segments and assign virality scores.
2. Visual Processing: It employs face detection to automatically crop videos into a 9:16 vertical format, keeping the speaker centered.
3. Enhancement: The system generates AI-written hook titles, adds word-synced animated subtitles, and can optionally overlay Pexels stock B-roll footage and transitions.
4. Editing: A built-in editor allows users to trim, split, and merge clips before exporting them for platforms like TikTok, Reels, and Shorts.

Who it’s for

It is designed for content creators, podcasters, and streamers who want to automate the creation of short-form social media content while maintaining full control over their data and avoiding watermarks.

Highlights

• Self-hostable: Can be run on personal hardware via Docker Compose for unlimited usage.
• Flexible LLM Support: Compatible with major cloud providers (Google, OpenAI, Anthropic) and local models via Ollama.
• Smart Cropping: Automatic face-centered vertical cropping for 9:16 frames.
• Comprehensive Toolset: Includes word-synced subtitles, virality scoring, and an integrated clip editor.
• Multiple Access Points: Available as a hosted web app, iOS app, REST API, and an MCP server for integration with clients like Claude and Cursor.