Let’s get you
listening again.

Keep an internet connection while preparing languages. If Google is waiting for Wi-Fi, choose Use Mobile Data to approve a cellular download, or reconnect to Wi-Fi. Apple manages its own language downloads. Some setup stages do not provide a percentage; loading a model is separate from downloading it.

Open Language Model to see the available space and the actual error. Allow about 1.5 GB free for Small or 3 GB for Turbo during first setup. Saved models also need working space when loading. Free space in iPhone Settings → General → iPhone Storage, then retry. Do not delete Tony to make room unless you have exported the recordings you want to keep.

Speech recognition and translation support are different. Compare language support in Language Model shows Apple, Whisper Small and Turbo side by side. Translate only offers usable language choices; both speakers need speech support and a shared Apple or Google translation provider. Transcribe can offer more languages.

Check the speaking language, then check microphone permission in iPhone Settings → Apps → Tony. Apple speech may also need Speech Recognition permission. If Hold Screen to Talk is enabled, hold the screen while speaking. Move closer to the microphone and reduce background noise.

Return to Tony’s home screen, then open Settings → Language Model. Your active session keeps its current model. A complete downloaded Whisper model is reused when selected again; loading it can still require working storage. Repair Download is a separate action for a model that cannot load.

Choose Translate and select both languages. Prepare the speech and translation models for both directions. Once ready, tap the divider button or the other person’s side to change who is speaking. Each language keeps its own side of the screen.

Scroll back on a live printout to read earlier lines. History opens at the newest words and stays frozen while you browse; in Translate, switch between the two language tracks. Session history is not saved after leaving the session. To keep audio and text, use Record, then Stop. Saved items appear in Recordings, where you can play and share them.

Check the language, reduce background noise and speak near the microphone. High microphone sensitivity can pick up quieter speech but also more mistakes. Whisper Turbo may improve recognition, with higher storage and memory needs. Google’s offline translation is intended for everyday conversation; non-English pairs translate through English. Review important wording before relying on it.