🗣️ Let Gemini transcribe and translate conversations live

Record a conversation, and Gemini turns it into a clean transcript. Or use Google Translate to talk across 70+ languages in real time. Here’s how both work, what to try first and the privacy catch.

ChatGPT/Kim Komando

I need your help: Add Komando.com as a preferred source on Google

What’s new: Give Gemini an audio recording, and its new transcription model cleans the whole thing up for you: speakers labeled, filler words removed and mid-sentence corrections fixed.

Say, “Let’s meet Tuesday. Wait, actually, Wednesday,” and the transcript says Wednesday instead of preserving your entire thought process for future generations.

It works in more than 85 languages and can detect the language automatically. You can upload audio through Gemini on your phone or the web. Free accounts get shorter recordings, while paid Google AI plans can handle much longer ones.

Live Translate sits inside the normal Google Translate app and translates spoken conversations continuously in more than 70 languages. 

You talk. They talk. The app keeps translating instead of making everyone stop every six words to pass the phone back and forth like a tiny diplomatic summit.

Google says it tries to preserve the speaker’s rhythm and tone instead of turning every human on Earth into GPS Lady.

Think about how many conversations this could have changed for the better. The contractor explaining what he found behind your drywall. Your Uber driver overseas. A relative you’ve never been able to really talk with. The train announcement in Italy that causes every local to abruptly grab their suitcase while you look around in a panic. Or the neighbor you’ve waved to 400 times but never actually spoken with.

Suddenly, the language gap has a pretty decent bridge.

⏱️ The 60-second setup

To transcribe something: Record the conversation (use Voice Memos, Recorder or whatever is on your phone) > Open Gemini > Tap the + and upload the audio file > Prompt Gemini to transcribe/clean it:

Transcribe this recording, and label each speaker. Bold the most important points. Then give me five bullets explaining what I need to do next and when. Finally, list anything you weren’t confident you heard correctly.

I’d much rather have AI say, “I’m not sure what she said here” than confidently invent the name of a blood pressure medication.

To translate a live conversation: Open the Google Translate app > Live Translate > Pick both languages > Start talking.

If you don’t see Live Translate, update the app before assuming Google has personally excluded you. You don’t need a Gemini subscription for this. 

⚠️ The obvious: Tell people you’re recording them. Secretly recording a conversation is a great way to make a useful feature weird.

Live Translate isn’t a professional interpreter. Fantastic for casual conversations, travel and everyday situations, but get a qualified human interpreter for medical diagnoses and legal documents. Anything else where one wrong word matters, really.

Maybe you can finally graduate from smiling, nodding and hoping nobody just asked you a question.