Dictation

Dictate a message instead of typing it: your words appear as you speak, recognized on your device, in the language you choose.

Mac · iPhone · iPad

On this page

Courant's composer has its own dictation: your words appear in the message as you speak.

Dictate

  1. In a message, choose Dictate:
    • Mac: the Dictate button at the bottom of the message. Dictation goes into the text of the message.
    • iPhone and iPad: the Dictate button over the bottom of the message, or above the keyboard. When the subject is selected, dictation goes into the subject.
  2. Speak. The panel says Listening… and your words appear as you speak.
  3. Done when you've finished. Dictation also stops by itself after ten seconds without a word, or after ten minutes.

✕ (Cancel Dictation) stops and takes out what was dictated this time.

New lines

Say « new line » or « new paragraph » to start one. These English words work whatever the language; in French, « à la ligne », « nouvelle ligne », « nouveau paragraphe » and « point à la ligne » work too, and other languages have theirs (Spanish « nueva línea », « punto y aparte »…). Courant adds the spaces and capitals; it doesn't rewrite what you said.

Language

The panel's Dictation Language button shows the language Courant listens for; More Languages lists the others. Courant guesses it from, in this order: the language you chose for this conversation, the last one you chose, the language of the email you're answering, what's already written, Settings › General › Languages › Languages I read, then your device's language. A choice is remembered.

The first time you dictate in a language, the speech model may need to download (« Preparing dictation… »); it then stays on the device.

Permissions and privacy

The first time, macOS or iOS asks to let Courant use speech recognition, then the microphone. If you refused, Courant says so with a button to the system's settings.

Speech is recognized on your device: by Apple's speech model on Macs with Apple silicon and the iPhone 12 or later, by Apple's dictation model, also on the device, elsewhere. Only as a last resort, when neither can listen in your language, does the system's former recognizer take over, and it may go through Apple's servers.