Google's new tool cleans up your speech recordings automatically

Google has just released a new AI tool called Gemini 3.5 Transcribe that turns spoken words into written text. Think of it as a smarter version of the voice-to-text feature you might already use on your phone — but with some handy upgrades.

When you record yourself speaking, the new tool doesn't just write down every word exactly as you say it. Instead, it automatically cleans things up. Those little filler sounds we all make — "um," "uh," "like" — get edited out. The tool also understands specialized vocabulary from different fields, whether you're talking about medicine, law, or technology. And it works in more than 85 languages, not just English.

Google says this new transcription tool is a major step forward from their previous version. It makes fewer mistakes when converting speech to text, and it's better at understanding people speaking multiple languages. What's more, you can edit the text using only your voice — you don't have to type corrections by hand.

For people who record meetings, podcasts, interviews, or lectures, this could save a lot of time. Instead of manually cleaning up messy transcripts, the tool does much of the work for you. It's part of Google's Gemini family of AI tools (think of these as different specialized versions of Google's AI assistant), showing how the tech company continues to add new features to make AI more useful in everyday life.