Most obvious next
STT / pronunciation check
Across the board: tap mic, say the phonetic target, get a soft pass/retry. Built for real learners with varied accents — not a strict native-speaker gate.
Roadmap
Ideas on the board — nothing here is live yet. No speech recognition ships in this build.
Most obvious next
Across the board: tap mic, say the phonetic target, get a soft pass/retry. Built for real learners with varied accents — not a strict native-speaker gate.
The clearest next step: speak Thai into the mic across Study, Review, and Games. Matching should be forgiving of accents and partial attempts — score rhythm and tones gently, then nudge the phonetic target.
A unified weak/due queue that spans every selected lesson, with smarter caps so short sessions stay focused.
Optional cloud dialogue that still stays lesson-scoped — freer replies with safety rails, not open-ended chat.
Cache quiz and listen audio packs so Games work fully on the train.