Mandarin’s tones make the early stage of listening especially slow, because the same syllable in a different tone is a different word. The upside is that spoken Mandarin has no conjugations to untangle, and once the sounds settle the hours start to pay off quickly. Serval counts Mandarin; Cantonese is not identified separately.

How many hours of Chinese?

Learners who kept count have reported what they could follow at each stage. For an English speaker learning Chinese, roughly:

1. 100 hoursSlow, simple speech made for learners
2. 300 hoursLearner videos at normal speed
3. 600 hoursEasy native videos on familiar topics
4. 1,200 hoursMost everyday native speech, with effort
5. 2,000 hoursNative content feels comfortable
6. 3,000 hoursNearly everything, even fast speech

Based on the Dreaming Spanish roadmap. Chinese is in group 4 of 4 in the US Foreign Service Institute's ranking of how long languages take an English speaker to learn: among the furthest from English. So the landmarks are doubled, as learners of the furthest languages usually report. If you already speak a language close to Chinese, expect to need fewer. How many hours you really need.

How Serval recognises Chinese

Serval listens to short windows of each video's sound, up to thirty seconds at a time, and a small model on your own computer names the language. No audio is kept or sent anywhere. Chinese is one of the 28 languages we measured against real recordings, and every one was labelled correctly. A label is only shown once two independent checks agree; when the audio is not clear enough, the video's time is not counted rather than guessed.

Chinese here means Mandarin. Cantonese is not identified separately, so Cantonese videos may go unlabelled or be labelled Chinese.

You never have to tell Serval you are learning Chinese. You only mark the languages you already speak, and everything else counts, so you can learn Chinese alongside another language and see each one's hours separately.

Finding Chinese to listen to

Search YouTube for 中文 (普通话), Chinese's name for itself, together with a topic you already enjoy: cooking, games, news, travel. Native speakers searching in their own language find content that searches in English miss. Early on, look for videos made for learners, with slow speech and lots of pictures, and move to native videos as they get easier; the hour landmarks above tell you roughly when.

Chinese is written in Chinese characters, which you do not need to read to start listening. Serval goes by sound, not subtitles, so it counts the same whether you turn captions on or not.

Close relatives

Chinese belongs to the Sino-Tibetan group. Serval also identifies Burmese and Tibetan.

Questions

Does Serval work for Chinese?

Yes. Serval identifies Chinese from a video's audio, on your own computer, and counts the time you spend listening to it on YouTube, Twitch, Netflix and six more sites. Chinese is one of the 28 languages it was measured on against real recordings, all labelled correctly. You do not have to tell it you are learning Chinese; it works that out by listening.

How many hours does it take to understand Chinese?

As a rough guide for an English speaker, about 600 hours of listening for easy native videos on familiar topics, and about 2,000 for native content to feel comfortable. It varies from person to person, and a language close to one you already speak goes faster.

All 105 languages · How many hours of input · Immersion trackers compared