Open yulu. When prompted, allow microphone, reminders, and calendar access. Tap and hold the blue circle, speak in Chinese, English, or both, then release. yulu transcribes your speech, extracts the action, and lets you save it to Reminders / Calendar or share it to any iOS app via the share sheet.
yulu downloads WhisperKit's speech recognition model (~500 MB) on first launch. This happens once. After that, launches are instant.
Edit the Transcript field directly before extraction. The field is fully editable. Tap "Re-extract from edited transcript" once you've fixed any acoustic mishears.
Every field in the Extracted section is editable inline — Kind picker, Title, Date toggle, DatePicker, Notes. Adjust whatever is wrong before tapping Save or Share.
After extraction, tap "Share to…". This opens the iOS share sheet, which lists every app that can accept text. Pick your destination.
When the on-device model is uncertain about a complex sentence (e.g. holiday names, abstract time expressions), yulu can route the transcript to a cloud LLM for better accuracy. Two options:
Default is Off — yulu stays fully on-device unless you opt in.
In the History section, long-press the row → Delete.
Yes, when Provider is set to Off. WhisperKit and Apple Foundation Models both run on-device.
yulu registers an AppShortcut, so you can configure your iPhone 15 Pro+ Action Button to "Open yulu" via Settings → Action Button → Shortcut → Open yulu. Auto-record on launch is not yet supported.
Email [email protected] with a description of what happened, what you expected, and a screenshot if applicable.