What Google’s new voice AI wave is and why it matters
Google’s latest wave of voice AI upgrades is a set of connected features across Gemini, Google Voice Search, and Pixel that expand multilingual understanding, add automatic summaries of spoken conversations, and blend music recognition with conversation tracking into a single audio-aware system designed to follow spoken activity throughout your day. In practical terms, this means you can speak to the Gemini app in over 70 languages without switching settings, hold longer voice searches that become live chats, translate speech on the fly, and even capture call notes and transcripts automatically through AI note taking features in Google Voice. Pixel’s emerging Audio Memory concept hints at a future where your phone keeps a searchable history of the songs and conversations around you. Together, these updates turn Google Voice multilingual support from a niche convenience into a central way people interact with its services.
Gemini app voice input gets 70+ languages and code-switching
The Gemini app’s mic is now built for people who move between languages in one breath. Google says the Gemini app voice input feature supports “over 70 languages” and no longer needs manual language toggles. You can speak in Spanish, add English product names, or mix Hindi and English, and Gemini will keep up with the entire request. Android Authority tested mixed English–Hindi commands and found that Gemini understood them correctly, which suggests the recognition model is tuned for natural code-switching. The update is rolling out on Android and iOS now, with support for Gemini on the web following about a week later. Paired with Gemini Live Translate and the newer Gemini models in the app, this upgrade turns voice into a primary way to use Gemini for multilingual users who prefer speaking instead of typing.
Google Voice Search adds live chat, voice translation, and better song ID
Google Voice Search is changing from a quick one-shot query box into a mini control center for spoken interactions. The redesigned interface introduces a blue “auto search” pill that lets you turn automatic searches on or off, so pausing mid-sentence no longer forces an early search. Below that are mode buttons that flip Voice Search between standard Google Search, a Live mode that opens a Gemini Live-style conversation, a dedicated song recognition mode, and a Translate mode. This voice translation Google option listens to speech and translates it into another language, making it useful for travelers or bilingual chats. Moving song recognition into its own mode helps identify music in noisy places, instead of relying on the app to guess when a song is playing. All of this makes Voice Search worth considering again as an everyday entry point into Google’s AI tools.
Pixel Audio Memory and the next phase of song and conversation capture
On Pixel phones, Google appears to be turning the Now Playing song recognition feature into something broader called Audio Memory. Strings spotted in an Android System Intelligence update describe Audio Memory as a way to “keep track of what you hear throughout your day, from the music around you to your important conversations.” It keeps Now Playing’s ability to recognize songs on-device using a local database, with a “short digital fingerprint” sent to Google only when local matching fails, while promising that “background conversations and audio are never sent to Google.” New text mentions storing info about songs playing nearby or in apps in a music history, accessible from the lock screen or the Now Playing app. The conversation reference hints that Pixel could later tie recognized speech, meetings, or calls into a timeline of audio moments, sitting alongside music history.
AI note taking features turn calls and voice into searchable memory
Alongside these listening upgrades, Google Voice is gaining AI note taking features that turn phone calls into structured records. During a Voice call, tapping the “Notes” button starts recording and transcription while Gemini extracts key points and action items in the background. When the call ends, Google sends those AI-generated notes in the body of a Gmail message, and you can also open the Voice app to see the full transcript, audio recording, and summary next to call details. A recorded disclaimer tells everyone on the call that AI is capturing the conversation, and post-call notes and transcripts are only visible to the person who initiated the feature. New Google Voice users have this turned on by default, while existing users must enable it through Workspace Smart Feature Consent. This is an early example of AI turning spoken workflows into searchable, shareable documentation across Google services.
