Turn a long source video into a short hook opener, translated subtitles, and a merged remix.
Generate dubbed audio in another language using a local text to speech backend.
Create short clips from public comments on a source video for social media.
| insiderx-pro/logiccut | gao-ruilin/autorun | gu-cryptography/anykb | |
|---|---|---|---|
| Stars | 101 | 101 | 101 |
| Language | Python | Python | Python |
| Setup difficulty | hard | easy | moderate |
| Complexity | 4/5 | 2/5 | 4/5 |
| Audience | vibe coder | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
Requires an AI coding agent like Codex to set up and download models onto your machine.
LogicCut is an open-source video repurposing tool for content creators, designed primarily for Chinese-speaking users. The README is written in Chinese. The tool is built around a workflow where an AI coding agent, specifically OpenAI Codex, is given this repository and then executes the video processing pipeline by calling LogicCut's command-line tools. The intended workflow takes a source video (either a local file or an authorized link to YouTube or Bilibili) and produces a set of secondary content from it: a 15 to 30 second hook opener built around a theme from the original video, translated subtitles, an optional dubbed audio track in another language, video clips made from public comments on the original video, and a final merged remix combining all of these elements. All major processing steps run locally rather than on external servers. Speech-to-text transcription uses faster-whisper, which is a local AI speech recognition model. Speaker identification (telling apart multiple people in a conversation) uses a model called pyannote. For text-to-speech dubbing, the tool supports several local backends including RGAD Cross-Lingual TTS, FishAudio, IndexTTS2, and OmniVoice, with RGAD recommended as the default for Chinese translation use cases. Video downloading uses yt-dlp, and all video cutting and merging uses FFmpeg. The repository does not bundle model weights. When Codex sets up the project, it downloads the required models from GitHub or Hugging Face onto the user's machine based on which features are needed. Cloud AI APIs are optional and require the user to configure them explicitly, the default paths avoid requiring external API keys. Current status is public preview. Several features are labeled beta or experimental, including the highlight opener, video translation, and comment-to-video conversion. Some workflows require cookies to download video from platforms, which may be affected by platform policy changes. There is no web interface yet, the tool is CLI-first and designed for agent-driven use. The project is licensed under Apache 2.0.
A video repurposing tool for content creators that an AI coding agent operates, turning a source video into hooks, translated subtitles, dubbed audio, and remix clips.
Mainly Python. The stack also includes Python, faster-whisper, pyannote.
Use, copy, modify, and distribute freely, including commercially, as long as you include the license and note any changes.
Setup difficulty is rated hard, with roughly 1h+ to a first successful run.
Mainly vibe coder.
This repo across BitVibe Labs
Verify against the repo before relying on details.