Install guide

Prepare for app access and the guided caption setup.

Check release status, open XNN Translate when access is available, sign in if needed, choose source and language, and confirm preview.

  • Release status
  • Open app
  • Preview

First run

Install XNN Translate and reach your first preview.

XNN Translate is a Windows desktop application that runs on your own machine. After installing, the first run is a short setup: open the app, sign in, choose what it listens to and the language to translate into, then confirm the preview.

Download and run

Download the XNN Translate installer for Windows and run it to install the desktop app, then open it from the Start menu.

  • Download
  • Run installer
  • Open app

Sign in

Open the Account panel and Sign in. Activation binds the app to your machine so the on-device models can run.

  • Account
  • Sign in
  • Activate

Choose audio source and language

Pick the audio source you want captioned and set the Target language for translation. These are the two choices that drive the live captions.

  • Audio source
  • Target language
  • Start captions

Confirm the preview

Press Start captions and watch the Preview. When the translated captions track the speech, setup is complete and you can route the output to OBS.

  • Preview
  • Confirm captions
  • OBS setup
OBS setup

System requirements

Check that XNN Translate will run on your machine.

XNN Translate runs on your own hardware and uses a dedicated NVIDIA GPU for real-time captions, with a processor fallback when no usable GPU is present. Speech recognition and translation run on the device.

Minimum

Windows 10 or 11 (64-bit). An NVIDIA GPU with at least 1 GB of free video memory for GPU acceleration, or any modern processor for the CPU fallback. 8 GB of system memory. Several gigabytes of free disk space for the app and its on-device models.

  • Windows 10/11
  • 1 GB+ free VRAM
  • 8 GB RAM

Recommended

Windows 11 (64-bit). An NVIDIA GPU with 6 GB or more of free video memory so the higher-quality on-device models run in real time. 16 GB of system memory and an SSD with room for the model files.

  • NVIDIA GPU
  • 6 GB+ free VRAM
  • 16 GB RAM

Without a dedicated GPU

XNN Translate still runs on the processor alone. Captions are produced with a longer delay and lighter on-device models, so a dedicated GPU is recommended for live work.

  • CPU fallback
  • Higher latency
  • GPU recommended
Troubleshooting

Ready for translated captions?

Choose the next product action that matches your workflow.