Voice Typing: dictate into supported apps
Voice Typing turns speech into text and can apply a chosen cleanup, rewrite, translation, or literal profile before inserting the result into a supported app.
How It Works
On Device AI can type the resulting text into the active supported app. Speech recognition can use Apple STT, Whisper, or supported Parakeet models for dictation.
- On your Mac: Press your configured trigger key, speak, then release it. The app inserts the processed text into the active supported app.
- On your iPhone: Open the On Device AI Custom Keyboard in any app. Tap the microphone, and dictate straight into your messages.
Correction Modes
Choose a profile based on how much you want the app to change the recognized text:
- Cleanup: Corrects common transcription and punctuation issues while preserving your wording where possible.
- Rewrite: Reworks casual speech into a style you choose.
- Translate: Produces text in a different language when the selected model supports the request.
- Literal: Keeps the recognized speech with minimal rewriting.
Language switching
You can choose a correction profile for a language, such as translating recognized Spanish into English.
Instant Short Phrases
For short replies, a short-segment threshold can skip the selected rewrite step. This keeps quick dictation from using a model when you do not need one.
Provider Customization
Choose the local model or optional cloud provider used for a profile.
- Use a compatible local model when you want the rewrite to remain on device.
- Switch to an optional cloud provider only after configuring it and reviewing its data practices.