Hold a hotkey, say the sentence, and the finished text lands where your cursor already is.
Mail, Slack, Notes, Pages, a browser field, Xcode, the terminal — SayToType inserts into the app you are working in, so there is no window to switch to and nothing to copy and paste.
Free to start. macOS 11 or newer, Intel and Apple Silicon.
Four steps between installing and dictating.
The Mac app is a universal .dmg that runs on both Intel and Apple Silicon, on macOS 11 or newer. Onboarding walks you through the two permissions macOS requires: Microphone, to record you, and Accessibility, so a global hotkey works and text can be inserted for you.
The default is Fn+Control. If that clashes with something you already use, there are Control+Option, Control+Escape, Fn+Shift and Fn+Escape to choose from — the Fn key is available as a modifier on macOS, which is often the one combination nothing else has taken.
Press the hotkey, talk, release. The text is inserted at the cursor in the app you are already in. Press Escape mid-sentence and the recording is cancelled with nothing written.
A mode fixes the recognition language, an optional language to translate into, and a prompt that shapes the result — terse commit messages in one mode, full sentences with punctuation in another. Switch between them without opening settings.
The one thing the Mac app does that the Windows app cannot.
Local modes use on-device speech models. Once a model is downloaded, dictation needs no internet connection at all, and nothing leaves the machine.
Two free and four premium, trading speed against accuracy: a balanced multilingual model, an English-only one, and faster or more accurate variants above them.
Local modes need an M1, M2, M3 or newer chip. On an Intel Mac the cloud modes and your own API key are the way to run it.
A model that has gone unused is unloaded on its own, so it is not holding memory between the moments you actually dictate.
Cloud or local, the rest of the app is the same.
Insertion happens at the cursor wherever you are — browsers, messengers, documents, code editors, the terminal. No integration to set up per app.
Pin the language in a mode, or let it be detected. A mode can also carry an output language: when you stop recording, what you said is transcribed and then translated by a separate high-quality model before it reaches your cursor.
A mode can carry a prompt, so the same spoken sentence comes out as a formatted note in one mode and a one-line summary in another.
Connect OpenAI, Mistral AI, Deepgram or any OpenAI-compatible provider. Work run through your own key spends no minutes from your plan — you pay that provider at its own rates instead.
The same Mac app turns recordings you already have into transcripts with speaker labels, timestamps and SRT or VTT export.
Cloud processing happens in real time and keeps neither the audio nor the text, on free plans included; what the app saves stays on your Mac. Dictation history is a switch you can turn off entirely, and local modes never send anything anywhere at all.
On a PC as well? See how dictation works on Windows — the hotkeys and the setup differ.
macOS 11 or newer. The download is a universal build that runs on both Intel and Apple Silicon Macs.
macOS requires it for two things SayToType depends on: catching a global hotkey while another app is in front, and inserting the finished text at your cursor. Microphone permission is the other one, for recording. Onboarding prompts for both.
For dictation, on Apple Silicon, yes — local modes run on-device models and need no connection once the model has been downloaded. Cloud modes need a connection, and transcribing a recorded file is online either way.
Fn+Control by default. You can switch to Control+Option, Control+Escape, Fn+Shift or Fn+Escape if that combination conflicts with something else you use.
SayToType adds custom modes with prompts that shape the output, translation into another language when you finish speaking, 47 recognition languages, the option of running on your own API key, and file transcription with speaker labels in the same app. If plain dictation in one language is all you need, the built-in tool is already there and free.
There is an iOS app for dictation, in beta. File transcription is desktop only.
A universal build for macOS 11 and newer, on both Intel and Apple Silicon.
Free to start — every account gets 50 minutes of cloud transcription a month at no cost, and no card is needed to sign up.
