High quality offline speech to text software for Windows. 17 MB installer, no runtime dependencies, no subscription, no telemetry, compatible with airgapped environments.
Powered by Mistral AI’s Voxtral Realtime model, the engine delivers accurate transcription across 12 languages: Arabic, Chinese, Dutch, English, French, German, Hindi, Italian, Japanese, Korean, Portuguese, and Spanish.
The software runs entirely on your local hardware – no internet connection required, no data leaves your machine.
It supports a broad range of GPUs, from integrated
graphics to discrete cards. See the performance benchmarks to find out
how well your hardware is likely to perform.
The software includes three applications:
Download the latest version
Version history
You can evaluate the software free of charge for 30 days. To evaluate on a computer without internet access, use the form.
One-time payment, permanent licence – no subscription, no recurring fees.
Screenshots – a dozen screenshots of the desktop software.
Frequently asked questions – additional information about the software.
About the project – technical background for the curious.
Voxtral Realtime is a model published by Mistral AI under an open licence. This software is an independent product and is not affiliated with or endorsed by Mistral AI.