Taiwan Mandarin, auto English spacing, and zero audio upload.
A private voice dictation keyboard powered on-device by Whisper large-v3-turbo. Dictate in any app without an internet connection. Strict Taiwan Traditional Chinese, auto English-Chinese spacing, and zero audio or text upload to the cloud.
SayIt runs the large-v3-turbo Whisper model locally. On first launch, Apple CoreML compiles and optimizes the neural network for your device's Apple Neural Engine (ANE). This happens once, and subsequent starts are instantaneous.
iOS keyboard extensions run in an isolated sandbox without direct microphone access. Full Access is required solely to communicate with the container App via a local App Group to start local recording and paste transcribed text. No typing or audio data is ever transmitted over the network.
iOS displays this indicator when the main App is in background audio standby. Audio is only sampled when you actively press the microphone button on the keyboard. You can tap 'End Standby' on the keyboard or set an idle timeout (1m, 5m, 15m, 1h) in Settings to dismiss it automatically.
Open the SayIt container app, navigate to the 'Correction' tab, and add a replacement rule. SayIt will automatically apply your custom substitutions to future dictations.
Log in with the Apple ID used to purchase Pro, open SayIt > Settings, and tap 'Restore Purchases'. Your purchase will be verified with the App Store with no additional charge.
SayIt requires an iPhone with at least 6 GB of RAM to run Whisper large-v3-turbo locally (including iPhone 13 Pro series, iPhone 14 series, iPhone 15 series, iPhone 16 series, and newer).
Questions, bug reports, or feature requests — email us and we will get back to you. Please mention your device model and OS version so we can reproduce the problem.