Voice for macOS

Speak.
Anywhere on
your Mac.

Hold a hotkey and talk. Your words appear at the cursor, in any app, any window. The same voice directs Claude AI assistants running on your Mac, by name, in parallel. Transcription happens on the device. Nothing is sent to the cloud.

Download for macOS View Source

PRIVACY

100% On-Device

LICENSE

MIT Open Source

PLATFORM

macOS 11+

§ 01. Voice for your Mac

Hold a key. Speak. The words appear where you're typing.

An email draft, a Slack message, a Google Doc, a terminal, a code editor: anywhere on your Mac with a cursor. Hold the hotkey, say what you want to write, let go. Cadence transcribes it on your Mac with Whisper and types it where the cursor is. The audio never leaves the device.

Typing

80wpm

A fast typist, sustained

· Hands on the keyboard.
· Eyes locked to the screen.
· One window at a time.
· You stop to switch apps.

Speaking

150wpm

A normal conversation, sustained

·Hands free.
·Words appear in under 200ms, on your Mac.
·Works in every app, no plugins.
·Talk while you read or think.

What Cadence actually does.

Cadence is a Mac app that lets you talk to your computer and talk to your AI assistants the same way: hold a hotkey and speak. Speech is turned into text on your Mac using Whisper, so no audio leaves the device. The same voice can direct several Claude assistants running in parallel, each addressed by name. No model picker, no cloud transcription, no subscription. Built by Cadence Labs, open source under the MIT license, runs exclusively on Anthropic Claude (Opus, Sonnet, and Haiku).

Latency

<200ms

First transcript token, on-device on Apple Silicon.

Privacy

100% local

Audio never leaves the device. No cloud transcription.

Models

Claude only

Opus 4.7, Sonnet 4.6, Haiku 4.5. No router, no GPT.

License

MIT

Free, open source, no subscriptions.

How Cadence compares

Capability	Cadence	Apple Siri	Cloud dictation apps
On-device transcription	Yes (Whisper)	Partial	No
Multi-agent orchestration	Yes	No	No
System-wide voice command	Yes	Limited	App-scoped
Open source	MIT	No	No
Powered by	Anthropic Claude	Apple proprietary	Varies

See It In Action

How It Works

Loading demo...

Architecture

What's actually running on your Mac

Speech recognition, on your Mac

Whisper runs directly on the chip in your Mac, tuned for Apple Silicon. Words start appearing in under 200ms. The audio never touches the internet.

< 200ms local inference

⚡

The right Claude for the job

Cadence picks between Claude Opus, Sonnet, and Haiku for you, depending on what you've asked for. Hard reasoning gets the strong model. Quick tasks get the fast one.

Opus · Sonnet · Haiku

Auto-routed

◉

Knows what app you're in

Cadence sees which app is in front, what text you have selected, and what's on the clipboard. So "fix this" in your code editor and "fix this" in an email don't get treated the same way.

⌘

Works in every app

Cadence types into whatever app you're already using: Terminal, your browser, a code editor, Slack, Mail, Figma. No plugins to install. No copy-and-paste.

~8MB

Binary size

0ms

Hotkey latency

100%

Local by default

∞

Extensible

Download

Get Cadence

Cadence. Apple Silicon Latest

macOS 12.0+ • ARM64 (M1/M2/M3/M4)

Architecture Apple Silicon (M1/M2/M3/M4)

File -

Size -

Released -

↓ Download for Apple Silicon Release Notes

Note: Requires Accessibility and Input Monitoring permissions. First launch: Right-click → Open to bypass Gatekeeper.

Cadence. Intel Latest

macOS 12.0+ • x86_64 (Intel Macs)

Architecture Intel x86_64