artbyjazi/autoclip
An open-source tool that turns a long video into ready-to-post vertical clips with burned-in captions
AutoClip takes a YouTube link or a video file, transcribes the speech, uses an AI language model to spot the moments worth clipping, reframes the footage to vertical while following whoever is speaking, and burns in animated captions to produce platform-ready MP4s. It can run fully offline using Whisper for transcription and Ollama for the language model, or connect to an outside API key for better results. The project is still pre-1.0, with the reframing quality not yet checked against a fixed set of test videos.
What it does
- Feeding in a long video runs the whole pipeline automatically: transcribe the audio, use a language model to find highlight-worthy moments, crop the footage to vertical (9:16) while tracking the active speaker, burn in animated captions, and export a finished video file.
- A fully local mode uses the Whisper speech-to-text model and the Ollama local model runner, so no video or audio ever leaves the machine; when an external API key is used instead, only the transcript text is sent, never the video or audio itself.
- Video reframing detects scene cuts first and frames each shot separately, and highlight timing is looked up from measured word-by-word timestamps rather than asking the AI model to calculate numbers itself, since language models are unreliable at math but reliable at copying a visible value.
- Four caption styles (bold pop, karaoke-style fill, clean lower third, boxed) and three export aspect ratios are supported, with both a web interface and command-line tools covering the same functions.
- The pipeline has been checked end-to-end on real talking-head footage, but the formal quality bar for reframing -- no visible jitter, no cut-off faces, speaker visible at least 95% of the time -- has not yet been validated against a fixed set of three benchmark videos, which is required before the project reaches its first tagged release.
Why it matters
Creators and marketers can turn long-form video into short clips without uploading footage to any company's servers, paying for a subscription, or creating an account, which matters for anyone handling private or unreleased source video. Running entirely on local hardware also removes ongoing API costs for repeated clipping work.
Terms in this repo
- Whisper · an open-source speech-to-text model that converts spoken audio into text
- Ollama · a tool for running AI language models directly on your own computer
- LLM · large language model, an AI system that understands and generates text
- diarization · the process of figuring out who is speaking and when in audio with multiple speakers
- libass · a software component used to burn captions permanently into video
Repository description (English)
Open-source, local-first AI video clipper. Long video in, caption-burned speaker-tracked 9:16 clips out. Fully offline with Whisper + Ollama, or bring your own API key.
Open on GitHubTrending repos
- vorssaint/vorssaint-utilsOne free menu bar app replaces a dozen paid Mac utilities
- Alishahryar1/free-claude-codeA local proxy that lets coding AI agents run on 49 free or cheap model providers instead of one paid service
- freestylefly/awesome-gpt-image-2A library of 532 reverse-engineered prompts that turn GPT-Image2 into a predictable image-making tool
- block/buzzAn open-source workspace where humans and AI agents chat, code, and review in the same rooms
- NousResearch/hermes-agentNous Research's Hermes is an AI agent that gets smarter the more you use it
- virgiliojr94/book-to-skillA tool that turns technical book PDFs into on-demand reference skills for AI coding agents
- VoltAgent/awesome-agent-skillsA single hub collecting over 1000 'how-to' manuals that make AI coding assistants act like experts
- anthropics/claude-plugins-communityA shared shelf where anyone's Claude add-ons get listed for install