The listening drill that gives you nothing to choose from

Chinese Sentence Dictation

Every other listening exercise on this site plays one word. This one plays a whole sentence and asks you to write it down — in pinyin with an ordinary keyboard, or in characters if you type Chinese, with unlimited replays and every word marked. 900 sentences, all of them published somewhere else on this site.

How to Play
  • Listen, then type what you heard. Press the big speaker as many times as you need — replaying costs you nothing except a small first-listen bonus. Esc replays.
  • Set the speed you can actually hear. Five rates from 0.5× to full speed, starting at 0.75×. It re-times the same recording in your browser, so the voice and the tones are unchanged — there is just more room between the syllables. Every visit starts at 0.75× again.
  • Type in pinyin — no special setup needed. Just type the sounds you heard on an ordinary keyboard, tones optional (ni hao and nǐ hǎo both count), with a pinyin keyboard under the box for ü if you want it. Already have a Chinese input method? Switch to Characters mode to type hanzi directly instead.
  • Press Enter to check. Every character is marked against the sentence that was spoken, so you can see precisely which syllables went past you. Punctuation is never marked.
  • Read the reveal — nothing is timed. The sentence, its pinyin, its translation and a link to the page it is published on. Take as long as you like.
  • 8 sentences to a round. A perfect sentence scores 10, a run of them scores more, and a partly-right one still earns what you heard.
Only character mode goes on the leaderboard — the two modes ask for very different amounts of work per sentence.

Can you write down what you just heard?

No options, no multiple choice — just the sentence and an empty box.

Length
Type in
1 / 8 Sentence
0 Score
Streak
Length

Tone marks are optional — they are not marked, so hao, hǎo and hao3 all count. ü is the one that matters, because and nu are different syllables; you can also type it as v. To turn numbered pinyin into tone marks, use the tone converter.

Looks like you're typing the pinyin sounds rather than characters — that's fine! and it'll check your answer that way.
You typed:
Keyboard: Enter to check and to continue · Esc to replay
What a Sentence Looks Like

One from each length, written out in full — in the game you hear these rather than read them. Notice that the sentences are ordinary published lines rather than drill material: they were written for a dictionary entry, a story or a phrase page, and this exercise borrows them.

Short · 5 characters Dictionary example

我爱我的家。

Wǒ ài wǒ de jiā.

I love my family.

Medium · 10 characters Dictionary example

听到这个消息我很高兴。

Tīngdào zhège xiāoxi wǒ hěn gāoxìng.

I'm happy to hear this news.

Long · 13 characters Dictionary example

老师进教室的时候,大家都笑了。

Lǎoshī jìn jiàoshì de shíhou, dàjiā dōu xiào le.

When the teacher came into the classroom, everyone laughed.

Why Dictation Instead of a Listening Quiz?

A listening quiz gives you four things to choose between, and choosing is a much easier task than hearing. You can pass one by catching a single familiar word and ruling out the options that obviously do not fit — which is why a quiz score can stay high while real conversations stay impossible. Dictation removes the options. There is nothing to eliminate, nothing to infer from, and no way to be nearly right without hearing the words.

It shows you the gap

Marking is per character, so a sentence you got four fifths of tells you which fifth you did not hear. That is usually a grammatical particle, a measure word or an unstressed syllable — the parts a quiz never asks about because they are never the answer.

Replays are free

Listen as many times as you like, at any of five speeds from 0.5× up. It starts at 0.75× on purpose: native pace is one unbroken run of syllables until you have heard a lot of it, and a first sentence nobody can make out teaches nothing. A dictation you can only hear once measures memory rather than listening, and the point here is to keep going until you can hear it — not to be scored on your first impression of it.

Two keyboards, one exercise

Typing hanzi needs an input method, which most learners here do not have installed — so Pinyin, which just asks for the sound, is the default. There is a pinyin keyboard under the box for ü and the tone marks either way. If you pick Character mode and type plain letters, the page spots it and offers to move you back.

Nothing Here Was Written for a Game

Every sentence you hear is one the site already publishes somewhere a person can go and read it, and each one links back to that page. That is deliberate, and it is the reason there is no dictation file in this project: a made-up sentence sounds exactly like a real one, and in a dictation it is also the answer key. The 900 sentences here were chosen from 10,884 across these sections:

A sentence is left out rather than adapted whenever it cannot be dictated fairly — too long to be spoken, too short to be a sentence, too long to type, or missing part of its published reading. Nothing is shortened or repunctuated to make it fit, because the moment a line is edited it stops being a line this site publishes. The same corpus is searchable at the example sentence bank and played a different way at Cloze in Context.

Questions About This Game

Dictation is hearing a sentence and writing down every word of it. It is harder than a listening quiz because nothing is offered to choose between: you cannot pass by recognising one word and guessing the rest, so it exposes exactly which syllables you are not hearing. It is also the only listening exercise that forces you to hold a whole utterance in memory, which is the thing that fails first in a real conversation.

No. Pinyin is the default mode and it needs nothing but an ordinary keyboard: type the sounds you heard, tone marks optional, so ni hao and nǐ hǎo both count as right. A pinyin keyboard sits under the answer box for the one letter that does matter, ü.

Character mode is there for anyone who already has a Chinese input method installed, or who wants to paste hanzi directly, and you can switch to it any time. If you pick Character mode without an input method installed, the page notices the plain letters you are typing and offers to switch you back to Pinyin before it marks anything.

Neither. Punctuation is ignored on both sides because text-to-speech does not pronounce a comma, so marking it would test punctuation rather than listening. Tone marks are stripped in pinyin mode, so hao, hǎo and hao3 all count as the same answer — tones are a production skill and this site drills them where they can be heard, on the Tone Trainer, Tone Pair Arcade and the Tone Pitch Mirror. The one thing that is not collapsed is ü, because and nu are genuinely different syllables.

All of them are sentences this site already publishes somewhere else: hand-written dictionary examples, lines from the graded readers, phrases from the travel phrasebook and examples from the grammar library. Not one is machine-generated, and the build fails if a sentence appears here that cannot be found word for word in the section that published it. Every sentence links back to the page it lives on, so you can read the whole story or phrase page it came from.

As many times as you like, and yes. A dictation you can only hear once is a memory test rather than a listening one, so replays are unlimited and free — the clip is cached after the first play. There are five speeds, from 0.5× to full speed, and it starts at 0.75× because native-pace Mandarin arrives as one continuous run of syllables until you have heard a lot of it. Slowing down re-times that same recording in your browser without changing the voice or the tones, so the pronunciation at 0.5× is the pronunciation at 1× with more room between the syllables. A speed you pick stays for the rest of your session, and every visit starts at 0.75× again — so a rate you chose once for one hard sentence never quietly becomes your normal. Getting a sentence right on the first listen earns a small bonus, which is the only difference replaying makes.

By the number of Chinese characters in the sentence, not by HSK level, and that is deliberate. In dictation you have to hold the whole utterance in your head and reproduce every part of it, so a twenty-character HSK 1 sentence is harder than an eight-character HSK 5 one — an HSK picker would promise the opposite. Each sentence still shows its own HSK level when it is revealed.

It is completely free with no sign-up. Your best scores are stored in your browser on this device only. Submitting a score to the leaderboard is optional and uses a nickname, never a real name or email.
The Three Lengths
  • Short — up to 8 characters 300
  • Medium — 9–12 characters 300
  • Long — 13–30 characters 300
Length rather than HSK level, because dictation is bounded by how much you can hold in your head — see the FAQ.
Your Best (this device)