Skip to content

Local AI Live Translate v1.3.0

Choose a tag to compare

@github-actions github-actions released this 27 Sep 07:48
· 34 commits to main since this release

What's new in 1.3.0

Eight more languages. Portuguese, Italian, Turkish, Polish, Ukrainian, Dutch and Bengali can now be used for both the video and the subtitles. Cantonese can be used as a video language.

Language Speech engine Errors in tests (Google FLEURS)
Cantonese SenseVoice 4.7% of characters
Portuguese Omnilingual 14.3% of words (4.0% of characters)
Italian Omnilingual 13.4% (1.7%)
Dutch Omnilingual 22.7% (7.1%)
Ukrainian Omnilingual 25.7% (4.6%)
Polish Omnilingual 24.9% (4.1%)
Turkish Omnilingual 22.2% (4.0%)
Bengali Omnilingual 15.5% (7.2%)
  • The best engine is picked automatically. If you choose one that can't recognise the language, the popup says which one it switched to. Omnilingual is the default for the European languages and Turkish because it's 3–6× faster. Whisper-Small is a little more accurate for Italian and Dutch, but adds 1–2 s per line, and you can choose it on the Live tab.
  • Cantonese is shown in Traditional characters. Chinese subtitles for Cantonese are now translated into standard written Chinese (呢幾個字都表達唔到 → 這幾個字都表達不了), where before the characters were only converted. This also applies when Cantonese is auto-detected, and adds about 0.3–0.5 s per line. Cantonese isn't offered as a subtitle language, because the translation model writes formal Chinese rather than colloquial Cantonese.
  • Hy-MT2-7B is still the recommended translator. In a new 39-language-pair comparison, it matched Xiaomi MiLMMT-46-12B and Tencent's Hy-MT2-30B-A3B for quality and was the fastest of the three (485 ms per line).
  • Known limitations:
    • The original-language line from Omnilingual has no punctuation or capitals.
    • Urdu isn't included yet, because no engine recognises it well in Urdu script.

There are no new downloads: the new languages use the speech models already included. To update, replace the old folder with this one, or run git pull on a clone, then reload the extension in chrome://extensions.

Download

File What it is
Local-AI-Live-Translate-<version>-windows-x64.zip Everything, ready to run (Windows 10/11, 64-bit): server, speech models, a private Python and the Chrome extension. No installation and no internet needed to start it.
Local-AI-Live-Translate-extension-<version>.zip The Chrome extension on its own (it is also inside the full zip).
SHA256SUMS.txt Checksums to verify the downloads.

Get started

  1. Unzip Local-AI-Live-Translate-<version>-windows-x64.zip anywhere and run START_Local_AI_Live_Translate.bat.
  2. In Chrome, open chrome://extensions, turn on Developer mode, click Load unpacked and choose the extension folder.
  3. Start LM Studio's local server with a translation model (Tencent Hy-MT2-7B recommended), or use Ollama or Qwen Cloud.
  4. Open a video, click the extension icon, pick your languages and press Start Live Translate.

Running from a git clone instead? START_Local_AI_Live_Translate.bat downloads Python, its packages and the speech models on the first run (~1.1 GB, each file checked against a pinned checksum).

See the README for all the features and settings.

What's Changed

  • Credit LiveCaption by begin0808 as the project's origin by @xinjalin in #4
  • Add a Screenshots section to the README by @xinjalin in #5
  • Add Cantonese, Portuguese, Italian, Turkish, Polish, Ukrainian, Dutch and Bengali by @xinjalin in #6

Full Changelog: v1.2.1...v1.3.0