Vietnamese Pronunciation App · tone practice
Safe Vietnamese tone practice feedback: recording check first, one correction, then retry — not a vocabulary app or ASR transcription tool. Built for learners who know the word but still get misunderstood when speaking.
Pre-launch. No native-like or perfect-accent promise. Not medical or throat coaching. No guaranteed accent transformation.
The Problem
Most language tools treat pronunciation as an afterthought, using standard translation chatbot scripts or generic, game-like score indicators that say "Try Again" without telling you why.
You speak clearly, pronouncing every vowel correctly. Yet the coffee server or taxi driver pauses, looks confused, and switches immediately to English. The culprit is frequently a flat or incorrect tone contour shape.
Early deck: Northern (Hanoi) first — see Methodology.
Vietnamese tones are not just static accent marks on a flashcard. They are pitch movements over time. If your pitch fails to rise, break, or drop with the correct timing and shape, the word can land differently.
Generic pronunciation scoring apps give you a "75%" rating and leave you guessing. Đúng Chưa? isolates the signal trace, evaluates the pitch shape, and gives one retry cue: "Lift the end of the syllable — avoid dropping pitch early."
How it works
Not a chatbot conversation loop — hear → record → recording check → one correction → retry, one phrase at a time. Practice alone, no audience.
Hear the reference before the microphone opens. Clips are AI voices, labelled Northern or Southern. Until a clip is on the page, you hear a pitch-shape beep that is labelled as a guide, not a voice. This is not a native-speaker recording.
Say the word once. Safari and Chrome ask for the microphone. The pitch line is drawn after you stop, not while you are speaking.
Your pitch is drawn over the reference shape. A gap means the tracker lost the pitch. It is not a picture of a voice break, and it is not a green check.
Get a single, clear cue translated from the signal error. Zero information overload. One correction, one retry.
Strategic positioning
Speech recognition can transcribe words while missing tone shape. Đúng Chưa? is a tone practice lab that stays honest when audio is not scoreable.
| Feature Category | Vocabulary Apps (Duolingo) | AI Speaking Tutors (Chatbots) | Đúng Chưa? Calibration App |
|---|---|---|---|
| Primary Purpose | Vocabulary drills & grammar charts | Simulated roleplay conversations | Focused tone contour practice |
| Speech Input | Lexical sentence repeating | Free-form conversational responses | Pitch shape audio calibration |
| Accuracy Feedback | Basic pass/fail speech indicators | Sentence syntax check summaries | Post-attempt pitch over the reference shape |
| Correction Guidance | None; displays correct spelling | Text grammar corrections | One specific correction cue |
| Dialect Control | Mixed spelling / broad exposure | Synthesized AI dialects | Standardized Northern reference targets |
Early Access
We are building and validating Đúng Chưa? alongside early testers. Join the pre-launch waitlist to receive updates when testing spots open.
What you get
Not included
Voice recording is strictly used to evaluate spoken pitch shapes. Early waitlist iterations will store voice samples for telemetry, algorithm debugging, and reference baseline calibration only with clear consent.
We'll reach out when early testing opens. Want to help us shape the initial practice deck? Tell us a bit about your struggles:
FAQ
Standard language apps focus heavily on vocabulary recall and grammatical syntax. Đúng Chưa? is built solely for voice pitch calibration. It shows your tone shape against reference traces and gives clear correction cues rather than spelling grids.
No. We do not use large conversational language models to judge your speaking attempt. The pass/fail scoring is entirely deterministic and calculated from voice signal processing. AI is only used as a translator to convert pitch errors into plain coaching instructions.
Northern (Hanoi) is the first set. Southern is on the practice page and is labelled Southern. Web references are AI voices or a labelled pitch-shape guide, not named human recordings. Central is not in this set.