Hear a real person say the word, not a robot
On the board and up for a vote. Every vote moves it up the list.
Tap a word you don't know and you hear a native speaker say it, not a synthetic voice. The model in your ear is the one you're actually trying to copy, accent and all, so you're not practising a version of the word nobody says.
Today's best AI voices, ElevenLabs and the rest, are very good at one thing: speech a native speaker can follow. A native ear glosses over a wrong vowel or a stress in the wrong place. Your ear can't yet, because it doesn't know which part was off. That's how plenty of conversation apps and pronunciation trainers in this space end up teaching an accent nobody actually has. We use AI for the boring parts, but the voice you copy should be a person's.
This might be a feature.
Votes are how this climbs the list, and we read every one.
Got a take on this one? Talk it through with us in r/Reelang.
Discuss on RedditSee everything we're building
This card is one of many on our public board: what's live, what's next, and the ideas we're still weighing.
The full roadmapGet the round-up
A short digest of what shipped and which new ideas landed, so you can weigh in and upvote. Pick how often.
No spam, unsubscribe anytime. We only email when something moved.