Skip to content

slohmaier/LangDetectSpeech

Repository files navigation

LangDetectSpeech - NVDA Addon

A lightweight NVDA addon that automatically detects the language of spoken text and switches to the appropriate synthesizer voice. Uses fast-langdetect (FastText-based) for efficient offline language detection across 176 languages.

Features:

  • Lightweight and fast language detection using FastText
  • Automatic voice switching based on detected language
  • Whitelist for considered languages
  • Configurable fallback languages

Tested synthesizers

Not working

  • CodeFactory Eloquence
  • Codefactory Vocalizer

Bundled binary dependencies

This addon bundles fasttext_pybind*.pyd from the fasttext-predict package (source, MIT licence). The .pyd files are taken unmodified from the official PyPI wheel.

Included builds

File Python NVDA version
fasttext_pybind.cp311-win32.pyd 3.11 32-bit up to 2025.x
fasttext_pybind.cp313-win_amd64.pyd 3.13 64-bit 2026.x

Rebuilding for a new Python version

pip download --no-deps --dest _deps `
    --python-version <X.Y> `
    --platform <win32|win_amd64> `
    --only-binary=:all: `
    fasttext-predict
Expand-Archive _deps\fasttext_predict-*.whl -DestinationPath _deps\extracted
Copy-Item _deps\extracted\fasttext_pybind.cpXY-<platform>.pyd addon\globalPlugins\fast_langdetect\

About

NVDA Addon to predict the language of every spoken text and change the spoken languge, if available.

Resources

License

Stars

Watchers

Forks

Packages

 
 
 

Contributors

Generated from nvdaaddons/AddonTemplate