Voice input
Dictate a draft with the recognition engine that suits your device.
Choose an engine
Open Me → Preferences to choose a speech recognition language and engine. Available engines depend on the language, device, and OS.
The app includes system speech recognition and on-device model options such as Whisper and Parakeet where supported. On-device Whisper requires Pro.
Record and review
Tap the microphone in a supported composer and grant microphone or speech recognition access when prompted. Speak, stop recording, then review the draft.
Correct names, numbers, and punctuation before sending. Dictation prepares text; it does not replace reviewing what you submit.
Prepare local models
Download the required model before using a local engine and allow enough storage. Preparation may show several files or phases rather than one overall percentage.
If recognition fails, verify permissions, model readiness, and language selection; try another available engine.