A Mandarin pronunciation test is the fastest honest answer to a question most learners cannot settle on their own: is it my tones, my consonants or my vowels that people struggle with? Mandarin is unusually hard to self-assess, because the thing that decides meaning most often, pitch, is the thing your ear is least trained to hear in your own voice. A scored test takes that judgment out of your ears.
This guide explains what a good Mandarin test measures, what the report tells you, which sounds learners most often lose points on, and what to practise for each kind of weak sound. You can try a few words right here. The full free Mandarin pronunciation test takes about three to five minutes: on a computer it runs in your browser with no account, and on a phone it opens in the TonePerfect app.
Short on time? Take the free Mandarin pronunciation test: 12 short prompts with pinyin and audio, an overall score and the sounds to practise first.
Why you cannot grade your own Mandarin
Three things make Mandarin pronunciation hard to judge from the inside.
Tones are part of the word. 妈 (mā, mother), 麻 (má, hemp), 马 (mǎ, horse) and 骂 (mà, to scold) differ only in pitch. If you grew up with a language that uses pitch for emotion rather than meaning, your brain files a wrong tone as "a slightly different way of saying the right word". You hear what you meant, not what you said.
Several consonants have no equivalent in English. z, c and s; zh, ch and sh; j, q and x are three families of sounds that an English speaker hears as two or three. Without feedback they collapse into whatever is closest in your first language.
Listeners are polite. Native speakers and voice assistants both use context to repair your errors. Being understood in a conversation does not mean a sound was right; it often means the other person did the work.
A test does not get tired, does not fill in the gaps and does not care about being polite. That is its whole value.
How a scored Mandarin test works
Every Mandarin syllable has three parts, and a useful test keeps them apart:
- The initial: the consonant at the start, such as the sh in 是 (shì).
- The final: everything after it, the vowel plus any nasal ending, such as the -ing in 听 (tīng).
- The tone: the pitch pattern over the whole syllable.
In the TonePerfect test you read 12 short prompts aloud, each shown with pinyin and with audio to help. Your recordings are compared with native speech, and the sounds the engine can match are sorted into those classes: a tone (such as tone 3), an initial (such as zh-), a final (such as -ang), or a tone pair (such as two third tones in a row). That is how the report can point at "Tone 3" or the "zh-" initial instead of giving one vague grade per word. You can stop after eight prompts and still get a report, but the full twelve cover more sounds.
The test checks pronunciation only. It does not test vocabulary, grammar or your HSK level.
What your score means
The report on the web has two parts:
- An overall score out of 100, based on the prompts you recorded. It is feedback on this attempt, not your language level.
- Up to three focus sounds: the sounds that scored lowest across your prompts and came up often enough to trust. Each one is named (for example "Tone 3", the "zh-" initial or the "-ang" final), with its own score out of 100, how many times it came up, and a tip. If your recordings did not contain enough of any one sound, the report says so instead of guessing.
The full breakdown into separate tone, initial and final results is part of the placement in the TonePerfect app, not the web report.
Here is what each kind of focus sound usually points to:
| Focus sound | What it measures | What usually goes wrong |
|---|---|---|
| A tone (Tone 1–4) | Whether the pitch moves the way that tone requires | Second and third tone confused, a first tone that sags, a fourth tone that does not fall far enough |
| An initial | Whether the starting consonant is the right one | z/zh or j/zh mixed up, missing aspiration on p, t, k, c, ch, q |
| A final | Whether the vowel and the nasal ending are right | -n vs -ng merged, ü said as u, the "buzzing" -i of zi or shi said as "ee" |
| A tone pair | Whether a tone changes as it should next to another | Two third tones in a row both said as full third tones |
A few rules for reading it well:
- One take is a sample, not a verdict. A noisy room, a clipped start or a mumble can pull the overall score down. If it surprises you, record again before you believe it.
- A focus sound is already a pattern. It is listed because it came up several times and stayed low, so it is worth a week of practice.
- If a tone is on the list, start there. For most learners, tone errors are the main reason a listener misunderstands them.
The sounds that cost learners the most points
Tones, especially two and three
The second tone rises; the third tone, in normal speech, stays low. Learners often say both as "some kind of rise", so 买 (mǎi, buy) and 埋 (mái, bury) become the same word. Two more habits show up again and again: a first tone that drifts down by the end of the syllable, and a fourth tone that starts too low to fall anywhere. Remember too that tones change in context: two third tones in a row become second + third, as in 你好 (ní hǎo). Our guide to third-tone sandhi covers when the written and spoken tones differ, and the neutral tone explains the light, toneless syllables in words like 谢谢 (xièxie).
z, c, s against zh, ch, sh (and j, q, x)
z, c and s are made with the tongue tip flat against the back of the upper front teeth. zh, ch and sh pull the tongue tip back and up toward the roof of the mouth. j, q and x press the middle of the tongue against the hard palate with the lips spread. Mixing these up is the classic initials problem, and it changes words: 四 (sì, four) and 是 (shì, to be) are a famous pair. Our zh, ch, sh vs j, q, x guide walks through each mouth position.
The "-i" that is not "ee"
In 字 (zì), 次 (cì) and 四 (sì), and again in 知 (zhī), 吃 (chī), 是 (shì) and 日 (rì), the letter i does not mean "ee". After z, c, s it is a buzzing continuation of the consonant with the tongue still forward; after zh, ch, sh and r it is the same buzz with the tongue still curled back. This is often called the apical vowel. Saying "see" for 四 or "shee" for 是 is one of the most common reasons a finals score drops. Compare zi and zhi, or si and shi, with audio.
ü, nasal endings and aspiration
ü (as in 绿 lǜ, green) is "ee" with tightly rounded lips; said as "oo", 绿 turns into 路 (lù, road). Hear the pair at lv. Final -n and -ng separate 晚 (wǎn, late) from 网 (wǎng, net). And p, t, k, c, ch, q need a clear puff of air, which is the only difference between 怕 (pà, to fear) and 爸 (bà, dad).
How to improve each kind of weak score
| If your focus sound is | Do this |
|---|---|
| A tone | Take one syllable through all four tones (mā má mǎ mà), exaggerate the contour slowly, then speed back up. Record, compare, repeat only the tone that scored worst. |
| An initial | Work in minimal pairs: sì/shì, zǎo/zhǎo, jī/zhī. Say each pair slowly while feeling where your tongue tip is, then record both. |
| A final | Isolate the final: -an/-ang, -en/-eng, u/ü, then the buzzing -i of zi/zhi. Use the pinyin chart to find neighbouring syllables with the same final. |
Ten minutes a day beats an hour on Sunday. Start each session with your first focus sound, and move to real phrases once the single syllable holds. On the practice page you can paste any Chinese text and get per-syllable feedback on it.
What a pronunciation test cannot tell you
A test scores the syllables it asked for. It does not judge whether a sentence sounds natural, whether you chose the right word or whether your rhythm in free conversation is comfortable to listen to. It tells you what is off, and the guides above help with how to fix it, but a patient human listener is still the best judge of real conversation. If you want the technical background, how AI grades Chinese pronunciation and our AI pronunciation checker guide go deeper.
Take the test, then fix one thing
The free Mandarin pronunciation test doubles as the Mandarin accent test: same 12 prompts, same report. If you searched for a checker, start at the Mandarin pronunciation checker. Take the test, write down your first focus sound, and spend a week on that sound alone. Then take it again. The TonePerfect app scores every word you practise syllable by syllable, with separate tone, initial and final results, so the weak sounds from your test become your daily drills.
Testing another language? The same approach works for every course: