If you tinker
Wire your voice to your machine. Open projects, kick off builds, run a script, all from commands you built, on your own local model.
Dictation is the floor.
Not the ceiling.
Talk, and clean text lands in any app, privately, on your machine.
Then your voice does more.
- Rewrite a selection
- Draft to clipboard
- Run a command you built
Free • Mac & Windows
↓
Meet Lyre
A lyre that won't lie to you.
It all runs on your machine.
- Hold a key, speak. Clean text lands in any app.
- Then it does more. Rewrite, draft, or run a command you built.
- Nothing leaves. The AI runs locally, on your machine.
Lyre is push-to-talk dictation: hold a key, speak, get clean text in any app, whether it's Slack, your editor, an email, or a terminal. That part just works. What most tools stop at, Lyre treats as the floor. The same voice can rewrite your selection, draft your reply, or run a command you built, and the AI doing it runs on your machine, so none of it has to leave.
See it work
Watch your voice do the work.
so um the thing about local dictation is it should just work everywhere you type, no window switching
The ceiling: the thing no one else does
You build the commands.
The AI just picks the right one.
Your voice can actually do things: open an app, run a build, or fire a script that does ten things at once. Say one word; your machine runs the sequence.
-
“open the folder folder”
open ~/work/folder -
“play query on YouTube”
open youtube.com/?q=query -
“start my project”
./scripts/project.sh -
“build and test”
npm run build && npm test
Lyre can only run commands you've already written. It can't invent one. Ever.
The power of “my voice runs my machine,” without the part that keeps you up at night.
And it's your text, too. Highlight anything, anywhere, and ask:
Change the tone “make this warmer”
Reformat “turn this into bullets”
Draft new “write this as an email”
Ask “what does this mean?”
Why it's safe to do all this
It runs on your machine.
Your words never have to leave. The whole pipeline runs on-device, so nothing you say is shipped to someone's server.
- Local speech-to-text. Transcribed on your machine.
- Local AI. Point it at your own Ollama or LM Studio model.
- Cloud is optional. Opt-in, labeled, off until you ask.
Start by just talking. Grow into a voice that runs your machine.
How it works
Three steps. One keypress.
-
1
Hold a hotkey and speak.
A small pill slides out at the edge of your screen, showing the app you're aimed at. Warm mic, no ramp-up.
-
2
Lyre figures out what you meant, locally.
Dictating? It transcribes and cleans up the text before inserting. Commanding? It matches your words to one of your saved commands and fills in the blanks.
-
3
It happens where you already are.
Text lands at your cursor. A rewrite swaps your selection. A draft hits your clipboard. A command runs. No window switch, no paste dance.
Who it's for
Same app. It grows with you.
If you just want to talk
Hold the key, dictate anywhere, get clean text. Private by default, nothing to configure. The power's there the day you want it.
Trust
Private by architecture, not by promise.
- Nothing leaves by default. Local transcription, rewriting, and routing.
- Nothing is stored unless you ask. History is opt-in and local.
- The AI can't invent a command. Shell actions confirm first.
- Cloud is opt-in and labeled. On only when you turn it on.
FAQ
Questions, answered.
Does my voice ever leave my machine?
No, not by default. Transcription and the AI cleanup both run on-device. Cloud is an opt-in option for lighter laptops, and it's labeled every time it's used, off until you turn it on.
Which platforms does it run on?
macOS and Windows, from one app.
Do I need my own AI model?
No. Lyre works with a recommended local model out of the box. If you already run Ollama or LM Studio, you can point Lyre at your own model instead.
Is it safe to let my voice run commands?
Yes. Lyre can only run commands you've already written yourself, it can't invent one. Shell actions wait behind a confirm step by default.
Does it work inside any app?
Yes. Hold the hotkey, speak, and the text lands at your cursor in whatever app is focused, your editor, a chat box, an email, a terminal.
What does it cost?
Lyre is free. Everything on-device costs nothing, no account, no subscription. The only thing you'd ever pay for is a cloud API, if you choose to use one, and that goes straight to the provider, not to us.
The person behind Lyre
Hi, I'm Chirag.
I build tools I wish existed or were free 😅. Lyre is one of them, a voice layer that keeps everything on your machine. If you want to see what else I'm working on, follow along, or just say hi, my door is open.