Today we’re introducing Gemini 3.5 Transcribe, our latest transcription model built for incredibly precise, smart dictation across your favorite apps and devices. Remember when traditional speech-to-text meant shouting over background noise, constantly hitting backspace to fix misspelled words, and manually deleting every "um" and "uh"? Those days are over.... Gemini 3.5 Transcribe isn't just dictation — it’s active intelligence with precise, context-aware speech-to-text support in 85+ languages. The model automatically filters out filler words, formats unstructured speech, and even pairs with your screen context to execute voice commands. Watch as Gemini 3.5 Transcribe removes filler words and uses multimodal capabilities to seamlessly turn messy voice input and local files into a polished email draft.show more

— Try it out in the @Geminiapp on macOS and Gboard on @Android — Build in the Gemini API via @googleaistudio, @antigravity, and the Gemini Enterprise Agent Platform (public preview) — Coming soon to @googlechrome and Gemini Enterprise for Customer Experience Learn more:

@GoogleAI can't wait to see how it handles different accents. that's always the tricky part!

Explain?

"Um... impressive" is exactly the kind of phrase this model would delete before you even finished saying it. 85+ languages and it still understood my accent better than most humans do.

“isn't just dictation — it’s” 👀

This is expected to be good, trying it today

Would the table entry have worked if the presenter didn’t move his cursor back into Gmail after asking the work to be done on the 3 pics?

Will gemini 3.5 pro comes or not?

Putting Transcribe in Gboard, macOS and Chrome is the smart bit. Nobody wants to open a separate app just to dictate.

back to back, crazy times in AI and modelling, great launch!

omg, finally

The introduction of Gemini 3.5 Transcribe marks a significant advancement in speech-to-text technology by addressing the inherent limitations of traditional acoustic models. By leveraging large language model architectures, this model can effectively mitigate the impact of background noise and handle disfluencies through contextual understanding rather than simple phonetic matching. This transition from signal-based transcription to context-aware semantic reconstruction is what enables such high precision across diverse environments.

@grok what these pitchers want from a NAGA

Dictation was never the hard part. Cleaning the text after was. Filler words stripped and 85+ languages is the useful bit for me. I talk one job out loud, get clean text back, and that text becomes the sub-step my agent actually runs. Voice in, task out.

去中心化算力成本优势明显

Context-aware transcription is a big step beyond basic speech-to-text. When voice can understand context and trigger actions, it becomes a real automation interface.

Conditioning audio transcription on active workspace state (open buffers, variable names, and agent thoughts) fundamentally solves the biggest pain point in voice coding: camelCase and obscure file path hallucinations. Generic STT models always struggle with domain-specific codebase jargon without that contextual prior.

Precise dictation is a nice upgrade

Multimodal context-aware transcription with real-time filler filtering looks super impressive. The progress on handling unstructured speech and noisy environments has been incredible to watch.

If you can speak normally, with pauses, filler words, and unfinished sentences, and it gives you back something clean and well-written, it changes the way you work completely. Speaking was already faster than typing. The only thing missing was for the machine to understand.

Dictation finally done right. Time saved beats money saved

filler word removal is nice, but the real test is messy multi speaker audio with names and jargon. that is where dictation usually falls apart 🎙️

Transcribe looking precise. Curious how it handles overlapping speakers.

And there goes @WisprFlow!

works great tbh

Eliminating filler words like "um" is straightforward. The true challenge lies in discerning substantive hesitations from mere noise. This distinction is critical for applications like deposition transcripts and medical documentation, where accurate interpretation is paramount.

seamlessly turn messy voice input into a polished email draft means your boss just got a typo-free apology for the meeting you missed

先看看复杂口音和嘈杂环境下，能不能真省掉返工