Say it.Keep going.

Your words, straight into your apps.
Hold a shortcut. Speak your mind. Release to paste.
Local voice dictation, quietly at home on Windows.

View on GitHub See it on a desktop

Preview · Private repository

Your voice stays on your PC.Download the speech model once.
Dictate locally, without sending audio to a cloud service.

Actual inSpeak Home window: ready-to-dictate status, Ctrl + Win shortcut, Parakeet v3 model, statistics and recent test dictations. — Actual app · Recreated widget · Test dictations · Click to enlarge

Local speech recognitionPolish & EnglishYour own shortcutsMade for Windows

A little room for your voice.

Keep the app you’re working in.
inSpeak brings the words to you.

A note for tomorrow

Monday, 5 October

A few things before we wrap up

The changes are ready. Just one thing to check.

Your next sentence starts here…

Back to the work that matters.

Draft · Saved on this device

Desktop illustrationRecreated widget · Example text

Watch the recording, release and text insertion. Your microphone stays off.

From voice to text.

A shortcut, a microphone,
and your usual app.

Speak. Release. Done.

Hold to record, release to paste. Choose your own keyboard or mouse shortcut, or use a toggle.

Recognition happens here.

Parakeet v3 runs locally after a one-time download. Your history and dictionary stay on this PC.

Polish. English. And more.

Automatic language detection and punctuation. One model, 25 European languages.

A widget within reach.

Audio levels while recording, status while processing. Keep it in one place or let it follow your cursor.

Your words.
Spelled your way.

Names, acronyms, project terms. Save the spelling you want, add pronunciation variants and set correction rules for specific apps.

“M. C. P.”MCP

Find that sentence again.

Search, copy or correct a past dictation. Your history stays on this PC.

Current inSpeak Home window with dictation status, statistics and recent activity. — Actual inSpeak app

Before your first dictation.

What do I need?

A Windows PC and a microphone. Setup guides you through choosing your input device, downloading the speech model and setting your shortcut.

Does it work offline?

Speech recognition runs locally. An internet connection is needed for the initial model download; you can then dictate without a cloud speech service.

Where does the text go?

Into the active application through the configured insertion method. You can review and copy your dictations from the local history.

Is it available to everyone?

inSpeak is in development. The GitHub link leads to a private repository and requires access. Public releases will come later.

Write with your voice.

inSpeak for Windows.

View on GitHubWindows · Repository access required