English speaking practice without the blank mind

ChatGPT voice and apps like Speak give you a partner to talk to. That helps once you can already speak, but it leaves the hardest part to you: thinking up what to say while working out how to say it, which is exactly when your mind goes blank. Here you practice English speaking with AI that has the original story in hand. The story gives you what to say, so all your effort goes into saying it in English, and because the AI has the original, it knows what you meant: it shows the native way to say your own sentence, points out what you left out, and scores you out of 100. Then you say it again and compare both takes side by side, so you see the progress in minutes instead of chatting for weeks and guessing.

No. The Clear sounds score only counts words a listener would struggle to understand, and there is no “percent native” score. You can keep your accent and still get full marks. If the AI misheard you, open “What we heard”, read your take as text and record it again.

Yes. Part 2 asks you to speak for up to two minutes from brief notes, which is very close to retelling. Pro includes Part 2 practice on people, places, events and objects: retell a model answer, see which points you dropped, then try again.

Yes, it is free English speaking practice to start. The story at the top of this page works without an account or card, with 4 corrected tries a day, which is two full rounds. A free account gives you 10 tries a day and starter stories from 3 scenarios.

Any level from about A2 up. This English speaking practice with AI lets you set the difficulty yourself: every story comes with notes at three levels, Detailed, Keywords, or My language (notes in your own language). Pick the one you need and switch anytime. Each sentence has the original text and a translation, and new words come with explanations, so you are never stuck on a word you don’t know.

Yes. This is English speaking practice online: it runs in your browser on a phone or a computer, so there is no app to install. You only need a microphone: allow it when your browser asks.

Your recording is sent to our speech recognition and AI providers only to create your feedback. We don’t keep it, and we never use it to train AI. Your scores and saved phrases stay in your account. Nothing goes public unless you choose to share it.

Pro costs $79 a year or $29 every 3 months, both with every scenario. We email you 3 days before each renewal, and you can cancel anytime in your account and keep Pro until the end of the period you paid for.