When you start a Gemini Live session, a captions button now appears in the top-right corner. Tapping it brings up a floating box that shows a text version of Gemini’s spoken responses. The captions appear in the middle of the screen and stay fixed in place. They do not show what you’re saying in real time, but that transcript can still be found at the end of the session.
Captions only appear when you’re using the full-screen Live interface. If you exit that screen during a session, the captions disappear, but they will return if you re-enter before the session ends. Once enabled, the captions setting stays active for future conversations, so you don’t need to turn it on each time.
A new ‘Caption preferences’ menu has also been added to Gemini’s settings. It’s located under the ‘Interrupt Live responses’ toggle and links to the system-level caption settings. This gives users a bit more control over how the captions appear.
Gemini Live now with captions. | Image credit — PhoneArena
Before this update, Gemini Live required the phone’s volume to be at a certain level. If your volume was too low, you couldn’t start a conversation at all. That limitation is now gone. You can speak your prompt out loud and simply read the reply, which can be helpful in quiet places or when you don’t have headphones.
The idea behind Gemini Live is to create a more natural voice conversation with AI, but there are clearly situations where hearing a spoken response isn’t ideal. Whether you’re in a quiet room or just prefer reading over listening, the new captions option gives users more flexibility.
This update may not seem major, but it addresses a real-world limitation. It shows that Google is continuing to adjust Gemini Live based on how people actually use it. If you don’t see the feature yet, force-stopping the Google app in your phone’s settings might help trigger the update.
In my opinion this is a practical change that improves usability. Gemini Live still has room to grow, but small updates like this help make it more convenient and accessible in everyday use.
