Show HN: Vibe-coded AI video clipper that runs in the browser (github.com)

A new web-based tool has been introduced that allows users to easily create short, ready-to-post video clips from longer content such as podcasts and presentations, all running directly in the browser using WebAssembly. Users can upload a video, and the system uses AI for transcription and analysis, producing 3-4 concise clips with captions and smart cropping based on the audio content. Notably, the entire codebase was generated largely by AI, demonstrating the rapid development capabilities of combining AI tools with specialized software development kits (SDKs) and APIs. This tool is significant for the AI/ML community as it showcases a practical application of AI in video processing, emphasizing client-side computing to eliminate server costs and reduce latency. Utilizing technologies like ElevenLabs Scribe, OpenAI’s Whisper, and Google’s Gemini, the tool effectively handles transcription and analysis, ensuring frame-accurate clipping without relying on less reliable timestamp generation from LLMs. Moreover, user confirmation for speaker mapping addresses common issues in automatic recognition, while non-destructive editing allows flexibility in video formats and templates without the need for reprocessing, making it a promising asset for content creators.