introducing Gemini 3.5 Transcribe: our most precise speech-to-text model yet, designed for intelligent voice interactions this new model: - seamlessly handles self-corrections - removes filler words to deliver clean, formatted text - understands your natural intent and speaking style - captures audio, even in noisy environments - supports 85+... languages, including regional accents and diverse dialects try it today in the Gemini API and in AI Studioshow more

yay

Lightning fast with incredible precision :)

Me reading that "Gemini 3.5" and thinking it's finally Pro, and nah🤦🏻

Let's goooo 🗣️🗣️

🚀🚀

brb transcribing everything

Go @aar0n_wade! Your multilingual skills always amaze me.

🗣️🎙️

Woohooo

If you don't have gemini3.5pro or gemini4pro, no one cares what you do with antigravity.

we just want 3.5 pro, show us what your ai devision is capable of, you are not even the best in things that are efficient, GLM 5.3 flash and GPT luna are sooo much better

Speed and accuracy!

I almost thought they released 3.5 Pro

so where is 3.5 pro?

Bring it to Google Meet immediately please!

Awesome! 👏

@grok what are the API costs for Gemini 3.5 transcribe?

Your press and talk in Gemini app and assistant sucks especially when I switch languagesat same sentence .. learn from chatGPT they are far ahead

the Chinese accent 👌

I have to admit your STT in the gemini app is insanely accurate.

regional accents getting treated like first-class input is long overdue

Good progress. Also wished for: - Text-to-speech function in Gemini app. - Both generating from text and text on image. - Several voices to chose from. Male / female. - Download as mp3.

A fun stress test would be a child retelling a story: false starts, sudden character changes, and a lot of “and then.” If the model preserves the idea while cleaning the transcript, that’s genuinely useful.

Gemini 3.5 Transcribe: our most precise speech-to-text model yet, designed for intelligent voice interactions

Thanks for giving us an early peek into this new model - it's a huge leap towards intelligent real-time transcription, and gives devs more control over what happens between conversations and the tool call. Our thoughts on Gemini 3.5 Transcribe Live:

The New Gemini 3.5 Transcribe is gonna go crazy with its own Ai powered smart glasses(yet to release) Y'all think this duo gonna beat the Meta ai glasses?

The self-correction handling is the bit I want to test :)

Just tried it in ai studio on a multilingual audio, where user speaks Arabic with some English words in between. English words were written in Arabic or sometimes entirely deleted. Although the regular Gemini 3.5 model used to transcribe this stuff well.

This is pretty impressive

The introduction of Gemini 3.5 Transcribe marks a significant step in making voice interactions more seamless and natural. The ability to handle self-corrections and remove filler words will likely enhance the precision of automated transcription tasks. How do these improvements impact the latency of real-time speech processing?

Whats the pricing?

Wow, this is awesome. It could sweep past whisper flow, I really do not understand why Google with so much in Deepmind keeps stacking gadgets but ships far too little tools that people can actually use.

Gemini 3.5 Transcribe now removes filler words and cleans up self-corrections. Great for meeting notes. I’d still want a verbatim switch for interviews and research, because sometimes the “um” is part of the answer.

Add to YouTube asap