Speak into your workflow.
Live local Whisper previews, a configurable shortcut, and a copy fallback when automatic insertion is unavailable. Add vocabulary, snippets, and writing modes.
A CortexKinematics project Experimental
Meet Cortala.
Keep your thoughts flowing.
Turn spoken thoughts into text. Choose a local speech model or your own cloud provider, transcribe meetings, and build small routines you can run with your voice.
Free experimental download · Apple Silicon · macOS 26 or later
Not notarized by Apple. Model setup is required. Read the installation notes.
Built for your everyday words
One voice. A few useful possibilities.
CortalaYour words, in progressA little less typing
Start with a thought.
Tomorrow, I’ll send the project update, share the next steps, and make time for the idea I’ve been putting off.
Product illustration · no microphone is active
Give conversations a second look.
Record microphone or system audio, transcribe meetings, and choose optional translation, notes, and anonymous speaker labels. Check important details before relying on them.
A phrase for your routine.
Arrange supported app, website, search, and wait actions into a voice workflow. Share workflows as JSON; imported routines start disabled until you enable them.
Your model, your choice
Local when you want it.
Cloud when you choose it.
With a local speech model selected, recognition runs on your Mac. Cloud providers and optional AI cleanup or translation are separate choices and may send audio or text to the endpoint you configure.
Models are downloaded separately. The DMG includes the speech runtime, but not models, Ollama, API keys, or personal recordings. Provider charges, if any, are paid through your own account.
How Cortala handles data →Before your first recording
A small setup. Then your words.
- Install the app.
Open the DMG and drag Cortala into Applications. This experimental build is ad-hoc signed and not notarized. macOS may block its first launch; managed Macs may not permit it. Review Apple’s current guidance. Never disable Gatekeeper globally or ignore a malware warning.
- Choose a speech model.
Download or select a local model in the app, or configure your own cloud speech provider. No developer credentials are included. Larger local models need more memory and can take longer to start.
- Allow only what you use.
Microphone access is needed for dictation. Accessibility enables auto-paste. Screen & System Audio Recording access is needed for meeting audio from other apps. Start with a short, non-sensitive test.
Open to a better next version
Free software. Room to contribute.
Copyright © 2026 CortexKinematics.
Help shape what Cortala becomes. Try it in your daily workflow, share what could work better, or contribute a fix, a new feature, documentation, or a translation. Every thoughtful contribution makes the next version better for everyone.
Explore the source and build something together. Cortala’s original code is GPL-3.0-only; third-party components keep their own licenses. Read the source and release notes to get started.