Accessibility in MagicTranslate
MagicTranslate includes captions and reading tools for people who are Deaf or hard of hearing and for people with low vision. These features are independent of cross-app text capture and do not use the macOS Accessibility API.
- No Accessibility permission: the Mac App Store edition does not request or use it.
- Screen Recording: requested only after the user deliberately starts screenshot capture.
- System Audio Recording: requested only after the user deliberately starts Live Captions; captures system audio only — never the room, never the microphone (a rare fallback path may use the Screen Recording channel instead).
- Copy-then-translate: copy text in another app, then press the translation shortcut. No Accessibility permission is required.
Live Captions for system audio
On macOS 26 or later, Live Captions turn audio playing in any app into synchronized on-screen text. ScreenCaptureKit captures the system audio and Apple Speech performs transcription on the Mac.
- Audio languages: English, Simplified Chinese, Traditional Chinese, Japanese, Korean, French, German, and Spanish.
- Display modes: original only, bilingual, or translation only.
- Adjustable text: four caption sizes — 20, 26, 32, and 40 pt.
- Easy configuration: move the pointer over the caption bar — including while no audio is playing — to change audio language, translation language, display mode, and font size. The same controls are available from the menu bar.
- Position and export: drag the caption bar to reposition it; export a session as SRT or VTT, or copy its transcript.
Caption accuracy depends on audio clarity, background noise, overlapping speakers, accent, and the selected language. MagicTranslate captions are not intended for emergency, medical, or other safety-critical communication.
Screenshot OCR and spoken results
Screenshot OCR helps turn small, low-contrast, or unselectable text into editable text. Apple Vision performs recognition locally first. The original OCR text and its translation can be read aloud; the default speech service is the macOS system voice.
- The screenshot is captured only after the user starts the screenshot command and selects an area.
- Screen capture uses Screen Recording permission, not Accessibility permission.
- Original and translated text have separate read-aloud controls.
- Privacy Mode disables all network-facing services, keeping OCR, translation, and speech on-device.
On-device and optional cloud processing
Live-caption transcription and the default Vision OCR and system-speech paths run on the device. Optional third-party cloud translation, OCR, and speech services are disabled by default. They are used only after the user explicitly enters their own credentials, enables the service, and selects it; content then goes directly to that provider under its privacy policy.
See the Privacy Policy for the complete data-handling description.
Text capture
MagicTranslate uses copy-then-translate and contains no Accessibility-based text capture. Copy text in another app, then press the translation shortcut.
Support
Questions or accessibility feedback: contact@shujuf.com or the support form. We normally reply within two business days.