VIDEO & LIVE STREAMS
How to translate video audio on iPhone
Follow videos and live streams in another language, with translated captions alongside the original speech.
You do not need a subtitle file to follow spoken audio in another language. Live Translate uses an iOS system broadcast to receive available system audio, transcribe the speech, and show a translation in a floating Picture in Picture window.
This is live speech translation: it creates captions from the audio you are listening to. It does not translate text inside a video image or create a translated voice track.
What you need
- An iPhone or iPad running iOS or iPadOS 26.0 or later.
- Live Translate installed from the App Store.
- A supported spoken language and translation language, with the required Apple-managed language models prepared.
- A source app that makes its audio available to iOS screen broadcasting.
Set up translated video captions
- Choose your languages. In Live Translate, select the language spoken in the video and the language you want to read.
- Prepare the language models. Use the model preparation option in the app. The first download requires an internet connection. Wait until the required resources are ready.
- Start Picture in Picture. This opens the floating caption window that stays visible while you switch apps.
- Start the system broadcast. Use the broadcast control in Live Translate and confirm the iOS system prompt.
- Play your video. Open the source app and play audible speech. When audio is available to the broadcast, Live Translate transcribes and translates it on your device.
Choose how captions look
Use bilingual captions to compare the original speech with its translation. You can also show only the translation or only the original transcription. Picture in Picture lets you move the caption window while watching the source video.
Why some videos do not produce captions
Some protected media and some source apps do not expose audio to screen broadcasting. Live Translate cannot bypass these restrictions. Quiet speech, background music, overlapping voices, an incorrect spoken-language setting, or missing language resources can also affect transcription.
If captions stop, check that the broadcast is still active, the source audio is playing, and the chosen models are ready. Try another source with clear, unprotected speech to narrow down the issue.
Live captions are not a saved subtitle file
System-audio captions are intended for live viewing. The saved history and text/PDF export features belong to microphone translation sessions. For a conversation you want to review later, use microphone translation.
Where the processing happens
Speech recognition and translation run on your device. Audio and subtitle content are not uploaded to the developer’s servers. Models may require an initial download, and online videos still require their own network connection. See the offline translation guide for the distinction.