Skip to content

JimLiu /baocut

AI video agent that transcribes, dubs and cuts inside an editable timeline
Editing & audioCustom licenseEnglish README

About baocut

BaoCut is a desktop app where an AI agent does the video work: transcribing, subtitling, translating, dubbing and cutting. Every result lands in a timeline you can keep editing, and fixing one subtitle line re-renders only that line, not the whole video. It drives your own agent engine (Claude Code and Codex are validated end to end) and can run local models on Mac and Windows. It's source-available, not open source: free for making and selling videos with it, but rebranding it or offering it as a hosted service needs a paid license.

Best for: Creators who want an AI agent to subtitle, translate, dub and cut footage but still want to hand-edit every step.

What it does

  • Transcribes with local models (Whisper large-v3, Qwen3-ASR) or online APIs, with word-level timing
  • Dubs into another language: translate, clone or pick a voice, then lay the new track on the timeline
  • Cuts talking-head recordings by removing filler words, pauses and retakes without touching the source media
  • Makes videos from a brief with 24 scene templates and motion graphics written as code
  • You and the agent share one editor and one undo stack; the agent reports a receipt of every change
  • Bring your own providers (OpenAI, Anthropic, ElevenLabs and more) with a per-job cost ledger

Quickstart

Rewritten from the README. Check the repo for the latest steps.

  1. 1
    Download the macOS (Apple Silicon) or Windows installer from GitHub Releases.
  2. 2
    Install ffmpeg and ffprobe; media analysis, transcription prep and export need them.
  3. 3
    Sign in to an agent engine such as Claude Code or Codex CLI.
  4. 4
    Open a video in the app and tell the agent what to do, or drive it from your own agent through the BaoCut skill or MCP.
  5. 5
    To run from source instead: Node.js 22.18+, npm install, then npm run dev; Rust is needed for the editor and local transcription.

Models and languages

Models it works with

  • Claude
  • Whisper

Interface and docs

  • English
  • Chinese

Alternatives

Other editing & audio projects people compare with baocut.

Huanshere

VideoLingo

Translated subtitles and a dubbed voice track for any video

Editing & audio
  • Qwen3-ASR
  • Qwen3-ForcedAligner
  • ElevenLabs
  • +7
19k GitHub stars+139/30d7 days agoApache-2.0

jianchang512

pyvideotrans

Video translated, subtitled and dubbed into another language

Editing & audio
  • Faster-Whisper
  • WhisperX
  • Parakeet
  • +7
19k GitHub stars+53/30dyesterdayGPL-3.0

harry0703

MoneyPrinterTurbo

Writes a narrated short video from a topic: script, voiceover, footage, subtitles

Video framework
  • MiniMax H3
  • Seedance
  • Wan
  • +7
129k GitHub stars+1.3k/30dtodayMIT

RVC-Boss

GPT-SoVITS

Clones a voice from a 5-second clip and reads your text in it

Editing & audio
  • GPT-SoVITS v2Pro / v2ProPlus
  • GPT-SoVITS v4
  • GPT-SoVITS v3
  • +7
62k GitHub stars+277/30dtodayMIT

Featured on OpenMicroDrama

Do you maintain baocut? Add this badge to your README so English speakers can find our write-up.

Featured on OpenMicroDrama
markdown
[![Featured on OpenMicroDrama](https://openmicrodrama.com/badges/featured.svg)](https://openmicrodrama.com/projects/jimliu-baocut)

The badge links to this page. Want your description changed? Email support@openmicrodrama.com.

Reviewed Oct 8, 2026. Approved from the discovery queue 2026-10-08. BaoCut Community License 1.0 (source-available): free for personal, research and in-company use and for selling content made with it; selling, rebranding or hosting the software needs a commercial license. Windows packages are unsigned and GPU inference is untested on hardware. We write these descriptions ourselves. The repo's own docs are the final word.