Your prompts are too long to type.
Speak them instead - local voice dictation for Windows,
local by default, cloud optional.
Windows beta available now
Starter needs no account. 5,000 words/day, every Whisper model, and CUDA acceleration are included.Press Ctrl + Space to dictate
Developers, writers, and professionals who value speed and privacy.
Local Whisper runs on your Windows PC. Your recording does not need to travel to a cloud API before text appears.
set it once · use it anywhere
Set one global hotkey, then dictate in the app already in front of you. Tap-to-talk and push-to-talk keep the control out of your way.
Transcribing locally…
Finish speaking, then paste the completed transcription at your cursor in one step. Works in editors, browsers, chat apps, and IDEs.
CPU only
works on any Windows PC
GPU (CUDA)
several times faster*
Test both engines on your device with the built-in 10-second benchmark.
*depends on your hardware and selected model
Run locally on any CPU, or use CUDA on a compatible NVIDIA GPU for faster transcription. Pick the model and engine that fit your machine.
scroll to explore · click to jump
Upload latency. Cloud audio. Typing bottlenecks.
Cloud dictation has to send audio away first. PrivateTranscribe can transcribe on the machine you are already using.
Your raw voice should not leave your machine just to become text.
Speak messy prompts, notes, and bugs. Clean them up where you already work.
The speed proof is simple. No upload step, CUDA when available, and text pasted where your cursor already is.
Add names, technical terms, and jargon so PrivateTranscribe can nudge Whisper toward the words you actually use.
Transcripts are saved in local SQLite, not synced to our cloud. Search, copy, and reuse past dictations on your machine.
Whisper can auto-detect 99 languages and optionally translate supported speech to English.
Optionally clean up transcriptions with AI - fix grammar, format for email, or restructure for clarity. Works with OpenAI, Anthropic, Gemini, Groq, or local models.
Starter is available now with 5,000 private words per day. Pro Early Access is for testers who want unlimited dictation and the workflow features while we finish them.
Polish transcriptions after capture - fix grammar, format for email, or restructure for clarity. Use local models or your configured OpenAI, Anthropic, Gemini, or Groq provider.
Pro featureCreate reusable AI instructions for cleanup, formatting, and tone so dictated text fits the place you paste it.
Pro featurePrivateTranscribe learns from your edits. Misheard the same word twice? It auto-corrects from the third time onward.
Pro featureUses your active app and window title for smarter transcription hints. Works across editors, browsers, and apps - no extensions needed.
Pro featureTrigger keyboard shortcuts, open apps, and run workflows with voice commands.
Pro featurePrivate dictation you can try properly. Upgrade pressure only appears after Starter proves useful.
Enough private dictation to prove it works on your machine.
Runs on any Windows 10 or 11 PC.
NVIDIA CUDA acceleration is included, but optional.
For people who hit the daily word limit and want the workflow layer as it lands.