Clips Kitty: local AI video clipping for content creators
Experience automated highlight extraction, speaker-aware face tracking, and local captioning to convert long streams into vertical shorts. Clips Kitty, created by ColinGPT9, runs its AI pipeline on your Windows PC so footage never leaves the machine. It automates transcription, face-based 9:16 cropping, and burned-in captions while offering multi-language transcription and optional translation. The tool is aimed at streamers and social media managers who need privacy-first, automated clipping without cloud uploads.
What does Clips Kitty actually do for long-form video?
Clips Kitty analyzes long recordings and extracts short highlights using local large language models and faster-whisper transcription. It accepts local files and links from platforms such as YouTube, Twitch, and Kick, runs speaker-aware face tracking with YOLOv8 and TalkNet, and produces 9:16 vertical crops with word-synced burned-in captions. The pipeline also generates titles, descriptions, and hashtags automatically, which speeds the task of preparing short-form posts.
How much system impact should you expect during processing?
Processing runs locally on the user's Windows PC, combining an Electron front end with a Python backend that calls faster-whisper and Ollama. That means CPU and GPU load depend on the models you run; the app is built to keep work on-device rather than sending data to a remote service. You can choose local model options to balance speed versus accuracy, useful when running on consumer-grade hardware during long batch jobs.
Is Clips Kitty safe to use with private footage?
The app keeps video data local by design, which prevents automatic uploads to cloud services and supports community auditing because the project is open source under AGPL-3.0. It uses local LLMs and transcription tools, so sensitive recordings remain on your disk. The developer supplies the option to run models like Ollama locally, reducing external network calls during the AI pipeline.
Do you need technical knowledge to operate it correctly?
Basic workflows are automated, but advanced configuration requires some familiarity with model selection and local tooling. The interface batches clipping steps: import, transcribe, detect highlights, crop, caption, render, and optionally auto-post. Casual creators can run defaults, while power users can swap models or adjust face-tracking parameters. The project being open source also means technical users can inspect or modify pipeline scripts if needed.
A practical, privacy-focused clipping tool with an alpha development caveat
Clips Kitty suits streamers and social managers who need on-device highlight extraction and captioning without cloud dependencies. It assembles transcription, face-aware cropping, and caption rendering into a single local workflow, and it supports multiple languages. Expect active alpha development status and occasional instability as the trade-off for early access to this privacy-first pipeline.
