Windows
SnailText for Windows.
Voice typing in any text field. Audio never leaves your machine. About thirty seconds to install.
.exe installer · ~20 MB
No account needed to start. Free with the compact local models.
Three steps after install
- 1
Run the installer. Windows may show a SmartScreen warning the first time — click More info → Run anyway. We're working on full code-signing.
- 2
SnailText launches into a brief setup — pick a model size and let it download (~150 MB for the compact one).
- 3
Press Ctrl Space in any app and start talking.
System requirements
- Operating system
- Windows 10 or newer
- Architecture
- 64-bit
- RAM
- 4 GB minimum
- Disk space
- ~150 MB plus model files
WebView2 runtime auto-installed during setup if missing
On a Mac? Switch to macOS →
What's in the download
One hotkey. Everything else included.
-
Two recognition models, free forever
The compact Whisper models — Standard (148 MB) and Fast (75 MB) — handle most everyday dictation: code comments, Slack messages, emails, notes. Pro adds three larger Whisper models plus NVIDIA Parakeet TDT v3 for higher accuracy in 25+ European languages.
-
GPU-accelerated by default, CPU fallback always
Whisper models pick the fastest backend on your machine automatically — Metal on Apple Silicon, Vulkan on Windows GPUs (NVIDIA, AMD, Intel Arc). On a GPU, Whisper Standard transcribes a short phrase in under a second; the larger Pro models stay comfortable too. Without a GPU it falls back to CPU and still works — slower on the bigger models, fine on the compact ones. Parakeet TDT v3 is CPU-only by design and stays fast there because the model is built for it.
-
A global hotkey that works anywhere
One key combination invokes voice typing in any text field on your machine — browsers, messaging apps, text editors, IDEs, terminal. No per-app integration, no extension to install in each tool. Press, speak, paste.
-
Local-only by default
Audio is captured into RAM, transcribed by the model on your device, and discarded. Nothing uploads, nothing logs to a server, no API key required. Free is unlimited — no recording cap, no trial timer, no watermark. Pro starts at $7.49 / month and covers up to 3 devices on one subscription.
What happens after the download
About thirty seconds from .exe to first dictation
Run the .exe installer. Windows may show a SmartScreen warning the first time — click More info → Run anyway. We're working on full Windows code-signing; until that lands SmartScreen will keep flagging fresh installers from new publishers, even legit ones. The installer itself is small (~20 MB) and includes the WebView2 runtime if your machine doesn't have it yet.
After install, SnailText opens into a brief setup screen. Pick a recognition model size — Standard (148 MB) is the recommended free option and covers most everyday dictation. The model file downloads in the background while you finish setting up your hotkey and microphone. No account needed to start; no credit card asked.
Press Ctrl+Space anywhere in Windows and start talking. The pill at the bottom of your screen turns red while recording. Press the same hotkey again to finish — text appears in whatever field your cursor was in: a browser address bar, a Slack thread, your IDE, a Word document, anything that accepts keyboard input.
Windows questions
Common questions before you install
Why does Windows show a SmartScreen warning when I run the installer?
Does SnailText need administrator rights to install?
Will SnailText work on Windows 10, or do I need Windows 11?
Want more on the privacy story? Read Offline dictation — voice typing without the cloud for the architecture, GDPR/HIPAA implications, and which dictation apps actually run on-device in 2026.