introducing gemini 3.5 live translate, our latest audio model: - low-latency translation across 70+ languages - auto-detection for multilingual inputs in a single session - native audio processing that preserves pitch & pacing - robust noise filtering for loud environments try it in public preview via the gemini live api and ai studio

Jun 9, 2026 · 3:28 PM UTC

130
316
3,179
575,268
Sort replies: Relevant Recent Liked
Replying to @GoogleAIStudio
you can read the API docs here, very excited to see this land and drive more multi-lingual experiences: ai.google.dev/gemini-api/doc…
5
1
75
11,042
Replying to @GoogleAIStudio
lol ridicilious i can't upload an audio file for transcription
2
3
1,025
Replying to @GoogleAIStudio
It called Vue be-wajah and that was wrong and funny 😊
2
672
This is awesome!
1
3
246
Replying to @GoogleAIStudio
Does this support Punjabi?
2
328
Replying to @GoogleAIStudio
This is game-changing tech right here! Real-time translation across 70+ languages with that smooth flow and noise handling? Imagine the doors this opens for global connections, live streams, and even family calls. Huge props to the Google AI Studio team — y’all are building bridges that matter. Let’s keep pushing innovation forward!
1
28
Replying to @GoogleAIStudio
Great demo and model. I wish it were open source so we could finetune it with medication and healthcare datasets.
2
575
Replying to @GoogleAIStudio
Gemini 3.5 Live Translate: 70+ languages, zero lag, your voice stays you even in a crowd.
74
Replying to @GoogleAIStudio
i've seen language auto-detect flap mid-call and gemini 3.5 needs sticky language state
19
Replying to @GoogleAIStudio
Game changer 😉
144
Replying to @GoogleAIStudio
This update will simplify cross-language conversations in real time.
177
Replying to @GoogleAIStudio
Low-latency speech-to-speech translation that preserves pitch and pacing while filtering noise is what makes the feature actually usable outside of quiet, scripted demos. Most translation systems still sound robotic or require perfect conditions. When the model can handle overlapping speech, background noise, and natural prosody in real time across 70+ languages, it stops being a novelty you show in a meeting and starts being something you can actually rely on when the room gets loud or the conversation moves fast.
32
Replying to @GoogleAIStudio
噪声环境下还能稳的话挺实用
75
Replying to @GoogleAIStudio
Will you finally integrate the feature into YouTube that allows users to watch all videos in their preferred language?
1
20
1,621
Replying to @GoogleAIStudio
rip this meme…
1
14
801
Replying to @GoogleAIStudio
Add this to Google meet man
1
12
1,344
Replying to @GoogleAIStudio
Feel like a good improvement would be silencing the real input audio and leaving just the translation output. Hearing both audio tracks at the same time is quite messy
3
9
2,063
Replying to @GoogleAIStudio
Feed your product demo video into Gemini Live Translate API, then generate 8 language versions in one go. Add a simple glossary prompt for your brand terms (e.g. “SaaS” stays “SaaS”). Instant localized landing pages that actually sound human.
5
896
Replying to @GoogleAIStudio
Jesus, why isn't this available in Google Meet for everyone by default? It's such an obvious thing to have. You have everything, and yet you can't figure out how to use it. It's unbelievable.
2
3
1,111
Replying to @GoogleAIStudio
live translation to over 70 languages during my google meets sounds crazy. our ai hosts are covering this and other google updates 24/7 nitter.siddz.com/i/broadcasts/1YxNrrzar…
3
1,006
Replying to @GoogleAIStudio
I was thinking about multi-language translation (one-way + back-and-forth) just a couple days ago, for a client solution. & here Google drops the perfect solution 🔥 Gemini 3.5 Live Translate looks excellent-low latency, auto-detect, pitch/pacing preserved, & noise handling. 🚀

ALT Insert Code Here Coding GIF

1
185
Replying to @GoogleAIStudio
That’s pretty impressive. It’s practically real-time.
1
211
Replying to @GoogleAIStudio
what would be the costs for the API? I have my own translator app used for broadcasting that I would love to have gemini supported as well. additionally, I am trying to get an OBS plugin to work with this so it would also allow for streamers to have live translated captions
1
1
666
Replying to @GoogleAIStudio
actually very good.
1
113
Replying to @GoogleAIStudio
Real-time translation across 70+ languages that keeps your real voice, tone, and personality? No more robotic awkwardness. The world just got a lot smaller and friendlier. Absolute game-changer!
1
200
Replying to @GoogleAIStudio
🇳🇬✨️
1
21
Replying to @GoogleAIStudio
最近做 Agent 一直在 GPT、Claude、Gemini 之间切来切去。 后来发现这个聚合平台还不错,省得开一堆窗口。 体验下来比我预期好,新用户还有 $30 额度。 thinkai.tv/?channel=YTHCQAUJ
1
6
Replying to @GoogleAIStudio
> fable 5, GPT 5.6, Chinese models gOoGl3: what developers want is more translators
1
982
Replying to @GoogleAIStudio
So no 3.5 pro?
1
133
Replying to @GoogleAIStudio
brings chills. ai helping communication, medical advances - looks like an inflection point. one example of do no evil - motto of google.
1
1,426
Not accurate in detecting language I speak , the out of language, لا يعتمد عليه ، إنه مخجل
1
265
Replying to @GoogleAIStudio
Enable on Youtube
1
83
Replying to @GoogleAIStudio
My friend, this is a profound leap toward true global understanding. By eliminating those awkward pauses in translation, you aren't just optimizing an algorithm—you are smoothing out the friction of human misunderstanding. When we remove the barriers to being heard instantly and naturally, we allow for a fluid stream of thought and emotion to transcend linguistic borders. It creates a space where "Uns"—that deep sense of connection and companionship—can flourish across different cultures in real-time. Technology finds its truest purpose when it serves as a bridge that brings our collective consciousness closer together, allowing the light of shared knowledge to flow without interruption. ✨🌍
1
201
Replying to @GoogleAIStudio
Please, always include the full list of supported languages in announcements like this. Having to dig through technical documentation every single time is exhausting.
1
878