Google's Gemini 3.5 Live Translate streams speech across 70+ languages in near real-time, available in Google Translate apps and in private preview for Google Meet.
Google on Tuesday shipped a speech-to-speech translation model built to keep pace with actual conversation. Gemini 3.5 Live Translate supports more than 70 languages and thousands of possible pairings, replacing the stop-and-wait mechanics that have defined most real-time translation products for years.
The distinction matters more than it sounds. Traditional systems queue an entire utterance, process it, then speak the result. According to CNET, Gemini 3.5 processes audio as it arrives, producing translated speech within seconds while attempting to mirror the original speaker's pacing, intonation and pitch rather than substituting a generic synthetic voice.
How it works
The model auto-detects the language being spoken, so users do not need to configure source and target languages before a call begins. Thurrott notes that Google describes the core engineering trade-off explicitly: the system balances waiting for context to improve quality against translating immediately to stay in sync with the speaker. That is the problem turn-based systems never had to solve.
Google says the model handles noisy environments, overlapping voices and informal speech. The company lists customer support calls, guided tours, classrooms, ride-share vehicles and live broadcasts as target use cases, each a context where earlier translation products have routinely fallen short.
Where it runs
The model is live today in the Google Translate app on iOS and Android. Android users get a Listening mode that routes translated audio through the phone's earpiece rather than the built-in speaker, useful in shared spaces. CNET notes the feature works with or without headphones.
In Google Meet, Live Translation is currently in private preview for select Google Workspace customers, expanding language support from five to more than 70. Language detection is automatic, removing the manual selection step. No general availability date has been announced.
Developers can access the model through the Gemini Live API and Google AI Studio in public preview, per Thurrott, with stated use cases including live video dubbing and multilingual conferencing. That positioning frames the underlying model as infrastructure, not just a consumer feature.
The competitive context
This launch drops Google into contested territory. Microsoft has offered real-time translation in Teams for years, and OpenAI's voice mode in ChatGPT handles conversational speech across multiple languages. None of those products, however, pair language breadth with a continuous-streaming architecture at this scale. If the voice fidelity claim holds, meaning translated speech preserves a speaker's emotional register and not just the literal words, it represents a meaningful step beyond what artificial intelligence has delivered in consumer translation to date.
Developer API access may matter more long-term than the consumer launch itself. Google Translate's REST API, introduced in 2011, became the translation layer embedded inside thousands of third-party apps. A live audio equivalent could follow the same pattern, given how many communication platforms are actively looking to reduce friction in multilingual meetings. Whether adoption follows depends on what independent artificial intelligence review finds when the product is tested under real call-center conditions, not curated demos. No third-party benchmarks exist yet.
The next signal to watch is how quickly Google Meet's private preview opens to all Workspace tiers, and whether Gemini Live API pricing works for smaller developers building on top of it.
FAQ
What languages does Gemini 3.5 Live Translate support?
The model auto-detects and translates more than 70 languages, enabling thousands of language pairings within a single conversation without manual configuration.
How is this different from standard Google Translate?
Standard Google Translate processes text or short audio clips. Gemini 3.5 Live Translate streams continuously, generating output as the speaker talks rather than waiting for a full sentence to complete.
Is Gemini 3.5 Live Translate available to developers?
Yes, through the Gemini Live API and Google AI Studio in public preview. Stated use cases include live video dubbing and real-time multilingual conferencing integrations.
When will Google Meet users get live translation?
A private preview is running now for select Google Workspace customers. Google has not announced a general availability date.
