discover · compare · collections · requirements · contribute · feed · watching · stats · help with your tokens
stronghamjji/PersoDub · source
Snapshot 0 · 2026-11-04T12:47:28Z · JavaScript · AGPL-3.0
Content fingerprint
9d607e21fb74bbfaa304ea448a7e9c5366854415cdea93d63136d94fa859c959GitHub description at capture
Open-source alternative to ElevenLabs, HeyGen & Rask AI. Clone a voice, dub your video into another language and post it to YouTube Shorts, TikTok or Reels. Runs locally on your computer.
README at capture
<p align="center">
<img src="docs/images/icon.png" width="96" alt="PersoDub" />
</p>
<h1 align="center">PersoDub</h1>
<p align="center"><strong>Natural dubbing from the best open models.</strong></p>
<p align="center">
<a href="#get-started">Download</a> ·
<a href="#hear-the-difference">Hear the difference</a> ·
<a href="docs/usage.md">How to use it</a> ·
<a href="docs/faq.md">FAQ</a> ·
<a href="https://github.com/stronghamjji/PersoDub/issues">Report a problem</a>
</p>
<p align="center">
<a href="LICENSE"><img src="https://img.shields.io/badge/license-AGPL--3.0-blue.svg" alt="License" /></a>
<a href="docs/requirements.md"><img src="https://img.shields.io/badge/platform-macOS%20%7C%20Windows-lightgrey.svg" alt="Platform" /></a>
<a href="https://github.com/stronghamjji/PersoDub/releases"><img src="https://img.shields.io/github/v/release/stronghamjji/PersoDub?color=green" alt="Version" /></a>
<a href="docs/privacy.md"><img src="https://img.shields.io/badge/privacy-footage%20can%20stay%20local-brightgreen.svg" alt="Privacy" /></a>
</p>
<p align="center">
<a href="https://github.com/stronghamjji/PersoDub/releases/latest/download/PersoDub-1.0.3-arm64.dmg"><img src="https://img.shields.io/badge/Download-macOS%20(.dmg)-000000?style=for-the-badge&logo=apple&logoColor=white" alt="Download for macOS" /></a>
<a href="https://github.com/stronghamjji/PersoDub/releases/latest/download/PersoDub-Setup-1.0.3.exe"><img src="https://img.shields.io/badge/Download-Windows%20(.exe)-0078D4?style=for-the-badge&logo=windows&logoColor=white" alt="Download for Windows" /></a>
</p>
<p align="center">
<img src="docs/images/finished-screen.png" width="100%"
alt="A finished job, English into Spanish: the script line by line on the left, each line with its original and its translation and how well the dubbed line fits its slot; the player on the right showing the dubbed video; and a timeline underneath." />
</p>
<p align="center">Drop in a video. Clone the voice. Dub it. Fix any line. Post it to Shorts, TikTok or Reels.<br>Open-source, in each speaker's own voice, on your own computer: the ElevenLabs, HeyGen and Rask AI alternative you run yourself.</p>
---
## What it does
<table>
<tr>
<td width="50%" valign="top">
<img src="docs/images/new-project.png" alt="Before dubbing: the video just dropped in, the target languages to pick from, and the Start dubbing button" />
### Dub
One video in, one dubbed video out, in each speaker's own voice. 10 languages on your computer, 100+ with the Perso API, depending on the dubbing model.
</td>
<td width="50%" valign="top">
<img src="docs/images/erase-screen.png" alt="The erase screen: a box drawn over the burned-in subtitle, ready to erase" />
### Erase burned-in subtitles
Box the subtitles, press **Erase**, and dub the clean video.
</td>
</tr>
<tr>
<td width="50%" valign="top">
<img src="docs/images/burn-in-screen.png" alt="The Burn in subtitles screen: a video and its subtitle file, twelve looks to pick from, the subtitle shown on the picture" />
### Burn in subtitles
Put a subtitle file on a video in the look you choose.
</td>
<td width="50%" valign="top">
<img src="docs/images/dub-agent-answer.png" alt="The Dub Agent, asked to make lines 1 and 2 fit, shortens them and remakes only those voices" />
### Fix every line
Edit a line by hand, or tell the **Dub Agent** "make line 2 fit". Only that line is remade.
</td>
</tr>
</table>
## Hear the difference
Same clip, English → Korean, each tool cloning the speaker's voice. The third video plays VideoLingo, KrillinAI, Voice-Pro and VoiceStudio one after another, each named at the top.
<table>
<tr>
<td width="33%" align="center"><b>Original (English)</b></td>
<td width="33%" align="center"><b>PersoDub (Ours)</b></td>
<td width="33%" align="center"><b>Other tools</b></td>
</tr>
<tr>
<td width="33%">
https://github.com/user-attachments/assets/39589651-83fe-4673-91b4-fa078f0e523e
</td>
<td width="33%">
https://github.com/user-attachments/assets/c5b3e0a6-de2a-4026-bdcd-9b3c0f8d179a
</td>
<td width="33%">
https://github.com/user-attachments/assets/d6da4598-1f27-4eeb-8f23-5988a6582daa
</td>
</tr>
</table>
Listen for pacing: PersoDub fits the translation to the time instead of speeding the voice up, and plays a line faster (10% at most) only when it would run into the next one. The source files live in [docs/demo](docs/demo): [VideoLingo](https://github.com/Huanshere/VideoLingo), [KrillinAI](https://github.com/krillinai/KrillinAI), [Voice-Pro](https://github.com/abus-aikorea/voice-pro), [VoiceStudio](https://github.com/debpalash/VoiceStudio).
## Get started
1. Download for [macOS](https://github.com/stronghamjji/PersoDub/releases/latest/download/PersoDub-1.0.3-arm64.dmg) or [Windows](https://github.com/stronghamjji/PersoDub/releases/latest/download/PersoDub-Setup-1.0.3.exe).
2. Drop in a video, or paste a link.
3. Pick a language and press **Start dubbing**.
**Updating from 0.8.4 or earlier?** Download and install the latest version once. From 0.8.5 on, PersoDub updates itself.
Start dubbing uses this computer by default and needs no key. Speech-to-text uses Whisper unless you choose Perso API, and on a new install translation uses ChatGPT through your own sign-in. To keep everything on your computer, set **Translation**, under **Dubbing step settings**, to **Hunyuan** or **Gemma 3**. To dub with the Perso API instead, save a free [Perso API key](docs/configuration.md) and set **Choose a model** to **Perso API**.
First setup is under 1 GB. A Perso API dub downloads nothing more. The first dub on this computer downloads Python + PyTorch and the models: about 10 GB on a Mac, 17 GB on a Windows PC with an NVIDIA GPU, 9 GB without one. A local translator adds 1.6 to 10 GB and the subtitle eraser 4.4 to 9.3 GB, so keep 30 GB free, 40 GB on Windows with an NVIDIA GPU. [Full table](docs/requirements.md)
<details>
<summary>Install notes</summary>
**macOS** — Open the `.dmg` and drag **PersoDub** into Applications. It is signed and notarized, so it opens with a normal double-click, and it checks for updates on launch.
**Windows** — Run the `.exe`. It installs for your user account only. It is signed by ESTsoft Corp. A new release can still show SmartScreen's "Windows protected your PC" for a while: click **More info**, then **Run anyway**. [Details](INSTALL.md#windows).
A screen-by-screen walkthrough is in [docs/usage.md](docs/usage.md).
</details>
### Install with your AI assistant
Paste this into Claude Code, Codex or any coding agent.
```text
Install PersoDub on this computer and check that it works.
Repository: https://github.com/stronghamjji/PersoDub
1. Check my OS, CPU, memory, free disk and GPU. PersoDub needs an Apple Silicon Mac
or 64-bit Windows 10/11 and 30 GB of free disk (40 GB on Windows with an NVIDIA GPU);
16 GB of memory is recommended, and dubbing on this computer needs at least 8 GB.
If this computer falls short, say so and stop.
2. Install the latest release for my OS:
https://github.com/stronghamjji/PersoDub/releases/latest
On Windows, SmartScreen may warn about a new release: More info, then Run anyway.
3. Open PersoDub and wait for the first setup (under 1 GB).
4. Start dubbing uses this computer by default and needs no key. Ask me if I want
a free Perso API key, for better transcription or the Perso API. If yes, open
https://perso.ai/dubbing?utm_source=desktop_app_github&utm_medium=desktop-app&utm_campaign=desktop_app&utm_content=install_prompt
then open PersoDub's Settings so I can paste the key in myself. Never ask to see it.
For a Perso API dub, set Choose a model to Perso API.
5. Dub one video under a minute with me. Before a first dub on this computer, tell me
the download size (9-17 GB).
If I pick ChatGPT for translation, I sign in myself. A free account is enough.
6. Report the version, where projects are saved, and how to reopen the app.
Finish the setup, not just a plan. Tell me any step you could not do.
```
## Why PersoDub
- **Your footage can stay yours.** PersoDub dubs on this computer by default. With Whisper and a local translator, none of it leaves. The Perso API, if you choose it, uploads the video.
- **Fix any line after the dub.** Change the words, the speaker or the voice of one line and only that line is remade. Cut on the timeline, or tell the Dub Agent what to fix.
- **Fits the words to the time.** Other tools speed the audio up. On this computer, PersoDub fits the translation instead, and speeds a line up (10% at most) only when it would run into the next one.
- **Voice cloning, not narration.** Each speaker still sounds like themselves.
- **Everything else is kept.** Music and effects stay. The translated `.srt` comes with the video, or burn it in, in the look you pick with **Subtitle style**.
## Details
| | |
|---|---|
| **Requirements** | Apple Silicon Mac or 64-bit Windows · 16 GB memory recommended · 30 GB free disk, 40 GB on Windows with an NVIDIA GPU. [Full table](docs/requirements.md) |
| **Languages** | 10 on your computer, 100+ with the Perso API, depending on the dubbing model. [Full list](docs/languages.md) |
| **Translation** | A local model, or ChatGPT for free. No subscription. [Configuration](docs/configuration.md) |
| **Better quality** | A free Perso key: better transcription, less to download. [Configuration](docs/configuration.md) |
| **Privacy** | By default the video is dubbed on this computer. With Whisper and a local translator, it stays here; the Perso API, if you choose it, uploads it. Which buttons are pressed is counted, never what you type; turn it off in Settings. [What is sent](docs/privacy.md) |
| **How it works** | The Perso API, or on your computer: Mel-Band RoFormer or Demucs, Whisper, CAM++, a local LLM or ChatGPT, Qwen3-TTS. [Architecture](docs/architecture.md) · [Pipeline](docs/architecture.md#2-the-dubbing-pipeline) · [Comparison](docs/comparison.md) · [Roadmap](docs/roadmap.md) |
| **Build from source** | [INSTALL.md](INSTALL.md) · [Development](docs/development.md) · [Contributing and security](docs/contributing.md) |