After 20 years of translations, Google Translate can now fix your pronunciation

Google Translate officially hit its 20th anniversary this week, and while the service has grown from a small AI experiment into a global tool supporting 250 languages (95% of the world’s population), there has always been one major missing piece. Knowing the right words is only half the battle; knowing how to say them is what matters in a real-world conversation.

To celebrate two decades of progress, Google is finally launching its most-requested feature: a dedicated pronunciation training mode.

Real-time feedback for your broken pronunciation

We’ve all been there, trying to read a translated phrase off a screen while a local stares back in confusion. The new practice tool aims to end that bit of tourist anxiety.

Once you receive a translation, a new “Pronounce” button will appear. Tapping it invites you to speak the text aloud, aided by a phonetic representation. Google then uses AI to analyze your speech and provides instant, detailed feedback.

It doesn’t just give you a “pass/fail” grade, though; the app actually highlights the specific parts of words that need improvement, allowing you to refine your delivery until you nail it.

Powered by Gemini

While the pronunciation trainer is the headline feature of this 20-year celebration, it’s part of a much larger shift toward what Google calls “Live experiences.” By integrating Gemini models into Translate, the app is moving away from word-for-word text swaps and toward enabling fluid, real-time conversations that track context, nuance, and even local slang.

Availability

The new pronunciation practice tool is rolling out now on the Google Translate app for Android. To start, it supports English, Spanish, and Hindi for users in the U.S. and India.

Given how much of Translate’s mobile audience is using the app for actual language learning (nearly a third of all users, according to Google), expect this trainer to expand to more languages very quickly.