Offline Speech to Text for Windows
Kokukeyboard is offline speech to text for Windows: a small, always-on app that turns your voice into typed text in whatever window has focus - your email client, a code editor, a chat window, a legal document. Press a hotkey, speak, and your words appear at the cursor a moment later. It's voice typing for Windows built entirely around local Whisper dictation, so it works as dictation without internet once the model is on your machine - nothing is ever uploaded, so there is nothing to intercept in transit.
or download the installer directly (.exe)No card for the trial. 30-day refund if you buy and change your mind.
How the hotkey flow works
Kokukeyboard sits quietly in the Windows system tray. Press the hotkey - Ctrl+Shift+D
by default, changeable to anything you like in Settings - and it starts listening. Speak a sentence,
a paragraph, a whole email; there's no need to pause between words or wait for live captions to catch
up. Press the hotkey again to stop, and the entire thing you just said is typed into whatever field
had focus: a Word document, a browser address bar, a terminal, a chat app. There's no separate window
to copy text out of and no clipboard step to remember. The cursor is the destination.
Because it listens to your whole thought before transcribing, Kokukeyboard skips the live, word-by-word guessing that some voice typing tools attempt - guessing that then has to be corrected by hand while you're still talking. It transcribes once, after you've finished speaking, and pastes the result in a single clean pass.
Because it types at the level of the operating system rather than plugging into any one app, there's no list of "supported programs" to check first. If a program accepts typed text, Kokukeyboard can put your dictated words into it - an inbox, a spreadsheet cell, a search bar, a terminal prompt, a form on a website. You don't switch tools depending on what you're writing in; the same hotkey does the same thing everywhere.
Why local Whisper dictation matters
Most dictation software works by streaming your voice to a server somewhere and streaming text back. That has three consequences: your audio leaves your machine, you need a live internet connection for the tool to work at all, and somebody is paying for that server time - usually by charging you per minute, per month, or both.
Kokukeyboard is built differently. Transcription runs through OpenAI's Whisper model - open source, and here running entirely on your own CPU or GPU. No audio is uploaded, ever; there is no code path in the app that could do it even if it wanted to. That's what local Whisper dictation means in practice: your voice stays on your machine, the same way a word you never said out loud stays private.
Two things follow from that. First, privacy: dictation for medical notes, legal work, or anything else you wouldn't want on a server stays genuinely private, because it's never transmitted. Second, cost: because your own hardware does the work, there's no per-minute meter running anywhere - which is part of why Kokukeyboard can be sold once instead of rented monthly.
Dictation without internet, in 90+ languages
Once the Whisper model is downloaded - a one-time step during setup - Kokukeyboard needs no internet connection to transcribe. Dictation without internet isn't a fallback mode here; it's the normal way the app works, on a plane, in a basement office, or anywhere the Wi-Fi is unreliable. The only things that ever need a connection are the initial model download and an occasional, brief license check.
Whisper recognizes 90+ languages automatically, so you can dictate in whichever language you're thinking in - or switch mid-thought if you're bilingual - without changing a setting first. Pin a single language in Settings if you'd rather trade that flexibility for a small accuracy edge.
Who offline dictation is for
Anyone who writes more than they'd like to type: long emails, documentation, notes between meetings, messages in a language your keyboard layout wasn't built for. It suits people who move between several applications through the day, since the same hotkey works in all of them, and anyone who works somewhere the connection isn't guaranteed - a train, a remote job site, a flight - since dictation without internet doesn't depend on being online at the moment you speak.
What it costs, and how to start the free trial
Kokukeyboard is $14.99 once, covering two computers - a desktop and a laptop, say - with no subscription and no recurring charge. Every feature is free to try for 14 days with no card required, and if it's not for you after buying, there's a 30-day refund.
Download the installer, run it, and the trial starts the moment you enter an email - no account, no card. If you have questions about installation, licensing, or languages before you start, the FAQ covers them, and the homepage has the full pricing breakdown. Batch-transcribing existing audio files instead of dictating live? That's a separate tier - see batch transcription. Curious why a desktop app can charge once instead of monthly at all? Read why Kokukeyboard is a one-time payment. On a Mac? See offline dictation for Mac.
What you install
- Installer: KokukeyboardSetup.exe, about 83 MB, digitally signed by IMHOTEONS s.r.o. It installs the Kokukeyboard tray app for your Windows user only, into
%LOCALAPPDATA%\Programs, no admin rights needed. - Speech model: downloaded on first run (a few hundred MB to a few GB, depending on the model you choose) and stored per Windows user. It survives updates and uninstalling.
- Uninstall: Windows Settings > Apps > Kokukeyboard > Uninstall, or the Start menu entry "Uninstall Kokukeyboard".
- Requirements: Windows 10 or 11, 64-bit; no graphics card needed, the AI runs on your processor.
- Price: $14.99 once for 2 computers, after a free 14-day trial (no card). 30-day refund.
- Published by IMHOTEONS s.r.o., Czech Republic. Read the privacy policy, the refund policy and the FAQ.