Real-time speech translation has been a demo-stage feature for years. Google Meet is now trying to make it boring, which is exactly what it should be. The company has expanded live voice translation to 63 languages and, according to Chrome Unboxed, dropped the manual setup that held the first version back.
Meet has offered translated captions for a while. Then Google added actual voice-to-voice translation, where you hear a translated version of the speaker instead of reading subtitles. It was a decent step. But the first release came with rigid limits: a small handful of supported languages and configuration you had to do yourself before a call worked properly.
What changed
The headline is scale. Going from a short list to 63 languages covers far more of the real world, including many pairs that businesses actually need but rarely get from software. The second change matters just as much. Setup is gone. Nobody has to pick languages in advance or fiddle with settings, so the feature behaves like part of the call and not a special mode.
That second point is the one I’d watch. Translation tools usually fail on friction, not accuracy. If a meeting host has to prep a session, most won’t bother.
How it compares
Microsoft Teams offers live translated captions and has pushed interpreter features inside Teams Premium. Zoom has its own translated captions and AI Companion tools. Both are strongest on text. Voice translation at this language count, built into a free-to-try product, is a harder thing to match. Dedicated players like KUDO and Interprefy serve high-stakes events with human interpreters and will keep that niche, but they cost far more.
Google also has an edge nobody else can copy easily. It owns Translate, years of speech data, and the Gemini models behind its newer audio work. Meet is a natural place to cash that in.
Does it matter?
Yes, with caveats. For distributed teams, support operations, and anyone working across Europe, Asia, and Latin America, this is a practical upgrade. A founder can hold a call with a supplier in another language without hiring an interpreter.
Still, treat it carefully. Machine translation can stumble on idioms, technical jargon, and fast crosstalk. Latency also changes how a conversation feels, since people wait for the translated audio. For contracts, medical talks, or legal discussions, a human is still the safer choice.
For everyday meetings, though, the barrier just dropped a lot. That is a real shift, not noise, and rivals will have to answer it.
