FREE WINDOWS BETA
Install and set up Talkasaurus on a Windows PC
Install the free beta on a supported Windows PC, choose a local speech model, and test your microphone before your first dictation.
1. Check your PC
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.
Allow space for the 454.8 MB installer, installed app, model and download/repair caches. Hatchling adds approximately 148 MB plus a small speech detector; larger models need more space.
2. Download and install
Download the free 0.7.33 Windows x64 beta from the official download domain. The installer is not digitally signed and Windows may show an unknown-publisher warning. Keep Windows security protections enabled; if you are unsure about the file, stop and ask for help.
The published SHA-256 checksum lets you compare the file's bytes. A matching checksum alone does not authenticate a publisher. Run the installer on the PC you intend to use and follow the setup wizard.
Verify before installing
For an additional check before running setup, download the verification kit. Extract it, place the downloaded installer in the same folder, and run INSTALL-OR-REPAIR.cmd. The launcher checks the installer against the current official HTTPS release metadata before offering to start setup.
The kit includes instructions for checking the file without installing it. This verifies file integrity; the beta remains unsigned.
3. Choose a model and microphone
Start with Hatchling, approximately 148 MB, and the small speech detector. Dictation models download separately; text-to-speech voices are bundled. Setup tests a speech sample on the GPU before accepting a model. Choose your microphone and check that it works.
For a PC that stays offline, download the model and matching speech detector on a trusted computer with an internet connection. Transfer the files to the offline PC and use the local model import option. Missing or mismatched files still need repair or replacement.
4. Try a short dictation
Open a compatible text field, configure your shortcut, then speak a short English sentence. Hold-to-talk and start/stop modes are available. F9 in the website demo is a configured example. Finish recording, wait for transcription, and review the inserted text.
To try the read-aloud feature, select a short passage and use your configured reading shortcut. F8 in the demo is an example. Talkasaurus does not send a message or submit an AI prompt for you.
If setup does not finish
Check the microphone connection, model files, disk space and graphics driver. Try Hatchling if a larger model does not fit your hardware. Some protected or elevated fields prevent insertion; try a normal editable field.
Contact [email protected] with the app version and a brief description of the failed step. Avoid including audio, dictated or selected text, passwords, license strings, full paths or private documents. This guide does not promise a live diagnostic-upload service.
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.
Version 0.7.33 · 454.8 MB · Windows x64 · English · Unsigned beta