v1.2.5 · Windows 10/11 · installer size varies by build

Type with your voice.
Anywhere on Windows.

One keyboard shortcut. Works in any app — Word, Slack, VS Code, browser. Offline. Private.

Download for Windows Windows 10 & 11 only macOS coming soon
Free forever, online No account required Open source CLI
slack.com · #design-eng
design-eng Design × Engineering · 28 members
S
Sarah Lin
Quick — can you push the spec update before the standup? 🙏
M
Mateo Cruz
+1, the API contract changed last night.
BI@

Three steps. No learning curve.

Press a hotkey, speak naturally, watch your words appear wherever your cursor is. That's the whole thing.

01 / hotkey

Press the shortcut

Anywhere — in any text field, in any app. Works system-wide on Windows.

ctrl + + v
02 / speak

Talk normally

Pause, "uh", restart, no problem. The model cleans up filler and punctuates for you.

03 / done

Text appears

Inserted at your cursor — formatted, punctuated, in the language you spoke.

push the spec update_

If you can type into it, you can talk to it.

VoltType injects text at the OS level — no extension, no plugin, no integration. It just works.

S
Slack
W
Word
O
Outlook
VS
VS Code
C
Chrome
N
Notion
D
Discord
L
Linear
T
Teams
G
Gmail
X
X / Twitter
Terminal
+ everything else with a text field

Your voice never leaves your machine.

VoltType runs Whisper locally. There's no server to send audio to, no telemetry, no account, no analytics SDK shipped with the binary. We can't see what you say because we never get it.

  • Local Whisper, on-device.The Pro plan ships a quantized Whisper model that runs on your CPU or GPU. Audio is processed in RAM and discarded.
  • Zero telemetry.No analytics, no error reporting, no "anonymous usage stats". The binary is sandboxed from the network entirely in offline mode.
  • Open source CLI & protocol.Inspect what we send, what we don't. The transcription engine is wrapped around whisper.cpp.
your-pc · windows 11
🎤 Microphone
16kHz mono · in-memory buffer
Whisper.cpp (local)
small.en · 244 MB · ~120 ms RTF 0.3
🖱️ Active text field
SendInput → cursor position
no cloud · no account · no upload

Free forever. Pro when you need it.

Free works great if you've got a connection and reasonable usage. Pro is for the offline-first folks who type all day.

// Free
$0forever

Everything you need to replace your keyboard for casual dictation.

  • Online transcription (cloud Whisper)
  • 30 minutes / day of dictation
  • Custom hotkey, custom dictionary
  • 14 languages, auto-detect
  • Offline mode
  • Unlimited usage
Download free
★ recommended
// Pro
$7/ month

Local Whisper. Zero cloud. Type all day, on a plane, in a SCIF, offline.

  • 100% offline — local Whisper model
  • Unlimited dictation
  • Faster — runs on your GPU (CUDA, DirectML)
  • Choose model size: tiny · small · medium
  • 99 languages, code-mode (no auto-correct)
  • Cancel anytime, 30-day money back
Start 14-day Pro trial
// Champion
$20/ month

For power users who want the same product and want VoltType to keep shipping faster.

  • Everything in Pro
  • Champion supporter tier
  • Helps fund offline model polish and Windows releases
  • Cancel anytime, 30-day money back
Start Champion

Questions you probably have.

How is this different from Windows built-in dictation?
Windows dictation is fine for short notes, but accuracy drops sharply on technical jargon, names, and code. VoltType uses Whisper, which has dramatically better word error rates for noisy audio, accents, mixed-language input, and domain-specific vocabulary. It also lets you set a custom hotkey, edit the auto-replace dictionary, and (on Pro) run fully offline.
Does it work in apps that don't accept paste?
Yes. VoltType uses the Windows SendInput API to type characters individually, exactly as if you were typing them on the keyboard. So it works in legacy desktop apps, Citrix sessions, RDP, secure text fields — anywhere a keyboard works.
Will it slow down my computer?
In free (cloud) mode, idle CPU usage is <0.1%. In Pro offline mode, it depends on the model size you pick — small.en uses about 800 MB of RAM and runs at 0.3× real-time on a modern CPU, faster on GPU. We never load the model until you first press the hotkey.
What about my privacy on the free plan?
On free, audio is sent over TLS to our transcription endpoint, transcribed, and discarded. We don't store audio, transcripts, or train models on your data. That said, if privacy is non-negotiable for you, Pro is the answer — your audio literally never leaves your machine.
Is my voice recorded or sent to a server on Pro?
No. In Pro offline mode, VoltType uses Local Whisper on your computer, and your voice audio is processed on-device instead of being sent to our server. The privacy policy explains the cloud/free plan separately: Privacy Policy.
Do you store my dictated text?
No dictated text is stored on our servers. The desktop app may keep settings, vocabulary, snippets, and transcription history locally on your own computer so the product works better for you; you control that local data.
Is VoltType GDPR-friendly?
VoltType is built around data minimization: account email, plan status, usage counters for billing, and local-only app data where possible. You can request account deletion or data questions at [email protected]; see the Privacy Policy for the current details.
What is the refund policy?
We offer a full refund within 30 days of your first subscription payment if VoltType is not right for you. After 30 days, partial-month refunds are not provided; the full wording is in the Terms of Service.
What happens to my license if I stop paying?
You can cancel any time, and the subscription stays active until the end of the current billing period. After that, paid Pro/Champion features stop renewing; the terms page has the subscription language: Terms of Service.
Why $7? Wispr Flow charges $15.
Because we don't have a cloud bill on Pro — your machine does the inference. We only charge for the desktop app, model updates, and a small dev team. If you want to support the project beyond $7, Champion is the $20/mo supporter tier.