Offline voice to text and read aloud for Windows

Dictation that runs on your PC

Hold a configured shortcut to talk, then release it to finish. Start/stop mode is also available. The local speech model transcribes the recording, and Talkasaurus inserts the result into the active text field in a compatible app. Review the text before sending or submitting it.

Use English dictation for AI prompts, email drafts, messages and notes. Protected or elevated fields can prevent insertion; app compatibility and hardware testing are still limited.

Read selected text aloud

Select a passage and use the configured read-aloud shortcut. The bundled text-to-speech engine and voices run locally. The homepage has a prerecorded voice sample; the website does not request microphone access.

What works offline

After installation and model setup, speech-to-text and text-to-speech processing stay on your PC. The app does not save recordings or a transcript history. You normally use the internet to download the installer and dictation model; verified local model import is also available. Downloads and updates are separate network actions.

If you insert text into an online service, that service handles the text under its own practices. Website visits and installer/model downloads involve network requests.

Free beta, with clear limits

Version 0.7.33 is a free, unsigned Windows beta. Commercial pricing and payment terms have not been announced. There is no purchase button or paid-trial activation in this offer. Transcriptions can contain errors; no accuracy percentage or universal application compatibility is promised.

See the setup guide and practical workflows before downloading.

Supported Windows beta requirements

Windows 10 version 21H2 (build 19044) or newer, or Windows 11, on an x64 PC. English dictation, a microphone, and a compatible Vulkan GPU with a working graphics driver are required. Compatible AMD, NVIDIA and Intel graphics use the shared Vulkan interface; this does not mean every card has been tested. Windows ARM, macOS, Linux, phones and tablets are not supported by this beta.

Setup runs a local speech sample before accepting a model. Performance depends on the GPU, driver and chosen model.