The connection may be missing. The words need not be.
Remote areas do not always have mobile internet. Directions, accommodation questions and conversations on the road should not depend on reception. The app is being developed to recognise and translate speech directly on the device. Suitable installed iOS voices read the result aloud; without a local voice it appears as text. No cloud translation service is required.
Speak. Understand. Reply.
Select source and target languages directly and swap them with one button. Choose the interface language independently. Tap the speaker language to start, then Stop to finish, automatically after at most 20 seconds. Original text and translation remain visible. Text input is also available; recording does not continue unnoticed in the background.
Choose the model size.
Compact, Balanced and Quality profiles require approximately 1.3 to 2.5 GB, plus installation headroom. Apple delivers packs on request. All languages share the models. Once installed, speech recognition and translation are processed on the device and local voices provide speech output.
Prepare before travelling.
Models and voices must be fully available on the device before offline use. Installation and updates are separate from offline processing. Development targets iPhone and iPad with iOS/iPadOS 26 or later; the final device list will follow before release.
Automatic translations can be wrong. They do not replace qualified interpreting for safety-critical or binding decisions.
15 languages. One app.
The app is not yet published in the App Store and no publication date is fixed. The planned languages are English, Mandarin, Hindi, Spanish, Arabic, French, Bengali, Portuguese, Russian, Indonesian, Urdu, German, Japanese, Marathi and Vietnamese. Quality and available speech output differ by language and device.