The short answer

Vietnamese looks friendly because it is written in the Latin alphabet, and then turns out to be one of the hardest languages here to say. Six tones, two of them separated by a glottal break, plus vowel distinctions English does not make. Tone is the word, not the emphasis, so an app working from a transcript has discarded the thing you most need corrected before the tutor ever sees it. Here the model receives your audio directly.

Vietnamese is misleadingly welcoming. It is written in the Latin alphabet, there is no new script to learn, the grammar has no tenses to conjugate, no genders, no plurals and no cases.

And then it is one of the hardest languages in this app to actually pronounce.

The tones are the language

Six of them in the Northern standard, marked with diacritics that look like accents and are not. They are the word.

Two of them, hỏi and ngã, are genuinely difficult for English speakers because both rise and the distinction is a glottal break in the middle of the vowel. In the South they are often merged, so listening to Southern speech does not resolve the question, and learners spend a long time producing something that is neither.

Why a transcript cannot help you here

Speech recognition takes your audio, decides which word you most likely meant, and returns text. The pitch contour that determined the decision is thrown away at that point, because writing does not encode it.

So an app that transcribes you and hands the text to a language model has already discarded the only evidence that your tones were wrong. It can help with your grammar, and there is not much grammar in Vietnamese to help with.

Here the audio goes to the model itself and comes back as audio. Nothing transcribes you first, so the tutor can hear that you got the syllable right and the contour wrong, and say it back to you properly inside its own reply.

North and South are both real

The curriculum teaches Northern Vietnamese, which is the standard form and has the full six-tone system. The tutor roster deliberately spans the country anyway, because the differences are systematic and worth meeting early: in the South, v behaves like a y, d and gi merge, tr and ch fall together, and the tone system simplifies.

Choosing Northern as the teaching standard is a decision about consistency, not a judgement. If your Vietnamese is for Ho Chi Minh City, talk to Minh and Khoa and you will hear it.

What a session is

Unscripted conversation with something you can interrupt, backed by an authored CEFR syllabus of 340 items from A1.1 to C1, of which 114 are capabilities: concrete things you can do rather than grammar points. The tone patterns above come from 20 pronunciation patterns authored for Vietnamese specifically, most of them tonal.

Going to Vietnam

Travel Mode changes the app for the length of a trip: rehearsal before you go, one real-world mission a day while you are there set in the city you are in, and a debrief with your tutor afterwards about how it actually went.

Vietnam is a good case for the mission format because everyday transactions come round constantly and the tolerance for a foreigner attempting the language is high, but the tonal precision required to be understood is real. Rehearsing the exact exchange, doing it, then working out afterwards what went wrong is a much faster loop than guessing.

The five Vietnamese tutors

Every language has five, each from a different place and each sounding like it. You can switch whenever you like, and the app remembers where you are regardless of who you talk to.

LinhHanoi
Northern Vietnamese, the basis of the standard form, with clear six-tone distinctions.
MinhHo Chi Minh City
Southern Vietnamese, with softer tonal contours and a warm, open delivery.
AnHuế
Central Vietnamese, with the gentle, lilting intonation the old imperial capital is known for.
TâmDa Nang
Central Vietnamese, clearer and softer than the deep Huế tones just up the coast, with the relaxed cadence of a beach city.
KhoaCan Tho
Mekong Delta Southern Vietnamese, with the soft, melodic vowels and dropped consonants typical of the river country.

What English speakers usually find hard

These are drawn from the pronunciation patterns the tutor listens for in Vietnamese. Because it hears your voice rather than a transcript of it, they are things it can actually notice and work on with you.

Six tones, and the marks are not accents: Ma, mà, má, mả, mã and mạ are ghost, but, mother, tomb, horse and rice seedling. The diacritics are not stress or vowel-quality marks, they are the tone, and the tone is part of the word.
hỏi and ngã are the two that defeat people: Hỏi dips low and comes back up. Ngã rises too, but with an audible break in the middle, a brief closure of the vocal cords. Both sound like the sharp rising tone to an untrained ear, and Southern speakers often merge them, which makes it harder to work out what you are aiming at.
nặng is a fall plus a stop: The heavy tone drops sharply and ends with the vocal cords closing, which makes the syllable sound clipped and shorter. English speakers skip the closure and produce something that sounds like the plain falling tone instead.
Final consonants close and stay closed: Final p, t, c and ch are formed but never released. Tốt and học end with the mouth shut and no puff of air. English releases its final stops, and doing so here is immediately audible.
Even syllables, and no English stress: Vietnamese is syllable-timed: each syllable gets roughly equal duration and keeps its full vowel quality. Importing English stress-timing crushes the unstressed syllables and flattens the tones at the same time, which is how one habit produces two separate problems.