ライセンス・第三者ソフトウェア表記

VoxShift uses LMT-60 model choices such as 0.6B and 1.7B. Users may download selected model files into the local model cache.

Official page: https://huggingface.co/NiuTrans/LMT-60-0.6B

VoxShift uses Qwen3-TTS 0.6B / 1.7B model choices and includes generated setup/test voice assets in the app package.

Official page: https://huggingface.co/Qwen/Qwen3-TTS-12Hz-0.6B-Base

Commercial builds avoid bundling PyAV / FFmpeg payloads and use a no-PyAV audio path for live numpy audio.

Official page: https://github.com/SYSTRAN/faster-whisper

Used when the RNNoise option is available in the local runtime.

Official page: https://pypi.org/project/pyrnnoise/

The app vendors Lucide 1.14.0 under realtime_translator/web_static/vendor/lucide.

Official page: https://lucide.dev/

The Microsoft Store package does not bundle VB-CABLE or launch a third-party driver installer. Users are sent to VB-Audio official pages to review the donation/license terms.

Official page: https://vb-audio.com/Cable/

The package may include or load NVIDIA/CUDA runtime DLLs through PyTorch, torchaudio, ctranslate2, bitsandbytes, or related dependencies.

Official page: https://docs.nvidia.com/cuda/eula/index.html

Used by the desktop shell when available on Windows.

Official page: https://learn.microsoft.com/en-us/microsoft-edge/webview2/concepts/distribution

Used to build this static website; not part of the VoxShift Windows app runtime.

Official page: https://astro.build/