PrimeDictate 6.0: local dictation on a new foundation

Signed x64 and ARM64 installers on .NET 10, a more forgiving wake word, and the shared core our next app is built on.

Most of us think faster than we type. Voice should close that gap. But most dictation tools ask you to send everything you say to a cloud service, or they rewrite your document while you are still talking.

We built PrimeDictate for a different workflow. You speak, you watch your words appear, and the finished text lands where you were working. On your own machine, without the clipboard.

Today we shipped PrimeDictate 6.0.

How PrimeDictate works

PrimeDictate lives in the Windows tray. Press your hotkey, or say a wake phrase like "okay computer", and it starts listening.

A speech model running on your PC turns your voice into text. The live words show up in a small overlay, never in your document. When you finish, PrimeDictate types the final text once, into the window you started in, and only if that window is still in front.

That rule is the point. Live retyping and clipboard pastes are where desktop dictation usually breaks. PrimeDictate does neither.

What's new in 6.0

• Built on .NET 10. The x64 and ARM64 installers are self-contained and signed, so there is nothing extra to install.
• A more forgiving wake word. "Ok" and "okay" count as the same word, and "thanks" and "thank you" variants are recognized.
• Cleaner voice commands. More spoken variants of a command phrase are stripped before the text is typed.
• A new overlay and tray icon. The overlay has a new particle animation while you talk, and the tray icon was redrawn.
• Command-line access. --show and --workspace open the PrimeDictate window.

The change you won't see

The biggest change in 6.0 is underneath. PrimeDictate now builds on a shared core library, the same foundation our next app runs on.

That next app is a cross-platform PrimeDictate with meeting transcription that labels who said what. We are testing it on real calls now. It is not part of this release, and it will ship through the same channels when it is ready.

A new runtime, a new architecture and a new build pipeline at once earned a new major version.

Upgrading from 5.x

The installer keeps the same product identity, so it replaces 5.x instead of installing beside it. Your settings, models and dictation history stay where they are, in your Windows user folder.

Get PrimeDictate 6.0

• Direct download: the x64 installer or the ARM64 installer, both on the 6.0.0 release page.
• WinGet: winget install --id FlowDevs.PrimeDictate --exact
• Chocolatey: choco install primedictate -y (package page)

The WinGet and Chocolatey updates are in their package reviews. Until they are approved, those commands install 5.1.0, and your next upgrade brings you to 6.0.

On first run, pick a speech model. PrimeDictate downloads it for you, and your audio never has to leave your PC.

The code is public on GitHub, and there is more on our PrimeDictate page.

Where this goes next

Dictation is the first workflow. Meetings are next: a transcript with each speaker labeled, made on your own machine. If your team has work that still runs on retyping, copying and pasting, let's talk about it.

Frequently asked questions

Does PrimeDictate send my voice to the cloud?

No. Transcription runs on your PC with local model files, so the core path has no cloud dependency.

Will upgrading to 6.0 reset my settings or models?

No. They live in your Windows user folder, and the 6.0 installer replaces 5.x in place.

Is meeting transcription part of 6.0?

Not yet. It is part of the cross-platform app we are testing now, and it will ship when it is ready.

Does PrimeDictate run on ARM64 PCs?

Yes. There is a native ARM64 installer next to the x64 one.

Download PrimeDictate 6.0