Private dictation for Windows

No typing.Just speaking.On-device transcription.

Push-to-talk transcription that runs entirely on your PC.Your microphone audio is not uploaded.

$29 onceFree trial
Try Whisperstream freeWatch it work

You think faster than you type.

Typing caps you at 40 to 60 words a minute. You think and speak at roughly three times that. Whisperstream closes the gap without sending a syllable to the cloud.

Typing0 wpm
Speaking0 wpm
See it work

Watch it type what you say.

Punctuated text lands wherever your cursor is.

GmailSlackClaude Code

Dictated live. Transcribed on the CPU, typed in. No cloud round-trip.

Works in the apps you already use

Try it free on your PCNo account needed

Everything dictation needs.

Core dictation and local cleanup can run on your PC. No subscription. See all features.

AI cleanup

Cleans up what you said

You said

hey um can you help me with the the report by Friday, sorry Saturday

Cleaned up

Hey, can you help me with the report by Saturday?

A local model cuts filler and applies your spoken corrections.

Or use an enhancement provider you configure.

Your vocabulary

Custom dictionary

clawedClaude
sarah nwenSarah Nguyen
my sequelMySQL
next jsNext.js

Add names, jargon, and acronyms once.

Switches with your app

Write a custom AI prompt for each app or site, applied automatically.

Transcript history

Search past dictations and replay the audio. Encrypted on your device.

Lockdown Mode

Turns off Whisperstream cloud enhancement and protects local history.

Import audio files

Drop in a recording and transcribe it, all on your PC.

Global shortcuts

Configure any key. Hold to talk, or toggle for hands-free.

Works across apps

Outlook, VS Code, your browser. No add-in.

39 languagesEnglish, Spanish, French, German, Italian, Portuguese, Russian, Polish, Bulgarian, Croatian, Czech, Danish, Dutch, Estonian, Finnish, Greek, Hungarian, Latvian, Lithuanian, Maltese, Romanian, Slovak, Slovenian, Swedish, Ukrainian, Arabic, Cantonese, Chinese, Filipino, Hindi, Indonesian, Japanese, Korean, Macedonian, Malay, Persian, Thai, Turkish, Vietnamese

All running locally, with no cloud.

Works offline

On a flight, on a train, or when the Wi-Fi drops.

Download free for WindowsBrowse all 15 features
The privacy guarantee

We physically cannot access your audio.

Here's exactly what happens when you dictate:

  1. 1Your microphone captures audio.
  2. 2The local NVIDIA model converts it to text.
  3. 3The text is typed into your active window.
  4. 4The audio is discarded from RAM. Nothing is saved. Nothing is sent.

Core transcription makes no network call and does not upload your audio. Separate connections support license checks, updates, model downloads, voluntary feedback, Agent Speak's Claude Code workflow, and any online enhancement provider you configure. A cloud enhancement provider receives transcript text only after you enable and configure that path.

Voice stays on your PCNo account needed

Local isn't a policy. It's how the transcription works. Nothing to leak. Nothing to log. No data to sell.

Download WhisperstreamInstalls in minutes. Runs offline.

Why rent dictation?Own it.

$29once

One-time purchase. Yours forever.

  • One license for up to 2 PCs
  • All features included
  • No account required
  • 30-day money-back guarantee
Buy once - $29Try free firstNo card needed
Whisperstream
Cloud apps
Speech recognition
On your PC
Remote cloud servers
Internet required
Not for dictation
Always
Privacy
No cloud transcription path
Audio sent to cloud servers
Yearly cost
$0/year
$140+/year
Cost over 3 years
$29
$420+
Account required
No
Yes

Common questions

Own your dictation.$29 once.On-device transcription.

Free to try. No account.

Download free for Windows30-day money-back guarantee
Compare Whisperstream

Guides and comparisons