How to learn Chinese from podcasts
Get to about HSK 4, read along when you can, and stay with one show so its words come back.
We make Mandarin The Wild.
What the studies found
Listening needs most of the words. One study changed how many of the words in short spoken stories listeners could know. With 90% known, most understood the facts, but learners varied a lot; with 95% known, learners understood relatively well and varied much less. van Zeeland and Schmitt (2013), Lexical coverage in L1 and L2 listening comprehension: the same or different from reading comprehension?, Applied Linguistics 34(4). On that 95% figure the authors put adequate listening at 2,000–3,000 English word families, against 6,000–7,000 for the 98% usually asked of reading. Nation (2006), How large a vocabulary is needed for reading and listening?, Canadian Modern Language Review 63(1).
Listening teaches words, but more slowly than reading. Learners who met the same new words in lectures and in texts picked up fewer of them by listening than by reading, at every level, though the gap narrowed as their level rose. Vidal (2011), A comparison of the effects of reading and listening on incidental vocabulary acquisition, Language Learning 61(1). A word has to come back many times before it is known well. Webb (2007), The effects of repetition on vocabulary knowledge, Applied Linguistics 28(1).
Reading along helps. Over 13 weeks with ten graded readers, learners who read while listening to the audio gained the most listening fluency, more than those who only read or only listened. Chang and Millett (2014), The effect of extensive listening on developing L2 listening fluency: some hard evidence, ELT Journal 68(1).
我坐地铁的时候听中文播客。
podcast
Most of this research is on English. So we counted Chinese.
How much of a Chinese podcast do HSK words cover?
We took 204 stretches of Chinese podcast talk, 455,000 words in all, from the WenetSpeech corpus's recordings from podcast platforms (interviews, news, talk, commentary, read-aloud stories), and 304 stretches from its YouTube videos, most of them drama. We checked each stretch's words against the 2025 HSK list. Zhang et al. (2022), WenetSpeech: a 10000+ hours multi-domain Mandarin corpus for speech recognition, ICASSP 2022 (transcripts as corrected by Paraformer). HSK: the official site of the Chinese Proficiency Test.
| HSK words you know | Podcasts (median, middle half) | YouTube videos (median, middle half) |
|---|---|---|
| HSK 1 | 51% (45–58%) | 52% (46–55%) |
| HSK 1–2 | 63% (55–69%) | 63% (56–67%) |
| HSK 1–3 | 73% (65–79%) | 72% (66–76%) |
| HSK 1–4 | 81% (76–87%) | 80% (75–84%) |
| HSK 1–5 | 88% (83–92%) | 86% (81–89%) |
| HSK 1–6 | 92% (88–94%) | 90% (86–92%) |
Podcasts sit a little above video at the top levels, and they spread wider: a read-aloud story is easier than a news bulletin. Even all six levels leave about one word in twelve unknown, short of the 95% at which listeners in that study understood well. van Zeeland and Schmitt (2013), Lexical coverage in L1 and L2 listening comprehension: the same or different from reading comprehension?, Applied Linguistics 34(4). Pick the easier end first: slow, scripted shows for learners, then interviews and talk.
Podcasts talk like people talk
Beyond HSK 4, the words in the most podcast stretches are words of talk, not of textbooks: 呀 (in 80% of stretches), 嘛 (59%), 人家 (53%), 呃 (51%) and 啥 (42%), and words with the northern 儿 ending such as 事儿 (75%) and 点儿 (33%), which the list writes without it. Learn these early and a lot of a podcast gets easier at once.
One show beats random episodes
Shows repeat their words. We split each stretch in half: 26% of the words above HSK 4 in the second half had already come up in the first half, against 6% in the first half of a different podcast. A show's hosts, guests and topics keep coming back, and so do the words for them.
A plan that fits the evidence
- Get to about HSK 4 first. Below that, more than a quarter of the words are new. Graded stories in the Library come with audio in two voices, and the level test shows where you are.
- Read along when you can. Reading while listening built listening fluency best, so pick shows with transcripts, or videos with Chinese subtitles you can point at with Mandarin The Wild.
- Pick one show and stay with it. Its words come back, so each episode is easier than the last.
- Learn the talk words early. 呀, 嘛, 啥, 人家 and the 儿 ending are in most episodes.
- Save the words that keep coming back to your Backpack and review them.
How we counted
The talk is WenetSpeech's machine transcripts, as corrected by the Paraformer model. Zhang et al. (2022), WenetSpeech: a 10000+ hours multi-domain Mandarin corpus for speech recognition, ICASSP 2022 (transcripts as corrected by Paraformer). Recordings whose names carry a YouTube video id are counted as YouTube videos; the rest come from podcast platforms. We read 40 fixed 256 KB slices spread through the file and kept every stretch of one recording with at least 800 words. Machine transcripts make some mistakes and have no punctuation, so a few words are misheard. Words are cut by our engine and levelled on the 2025 HSK list the way the Library levels a story; names the dictionary only knows as names are left out. "Middle half" is the 25th to 75th percentile of stretches.
Questions
Can you learn Chinese from podcasts?
Yes, once you know most of the words you hear. Studies find listening teaches words more slowly than reading, and reading along with the audio helps most. In our count of 455,000 words of Chinese podcast talk, HSK 1–4 words are 81% of a typical stretch.
What level do I need for Chinese podcasts?
About HSK 4 for everyday talk. Below that more than a quarter of the words are new; in one listening study, learners understood spoken stories well once they knew 95% of the words.
Are Chinese podcasts harder than Chinese TV?
About the same by vocabulary: HSK 1–4 words are 81% of a podcast stretch in our count and 80% of a YouTube video stretch. Podcasts vary more, from easy read-aloud stories to news.
Sources
Checked on 14 October 2026. Prices and features change; tell us if one has.
- van Zeeland and Schmitt (2013), Lexical coverage in L1 and L2 listening comprehension: the same or different from reading comprehension?, Applied Linguistics 34(4)
- Nation (2006), How large a vocabulary is needed for reading and listening?, Canadian Modern Language Review 63(1)
- Vidal (2011), A comparison of the effects of reading and listening on incidental vocabulary acquisition, Language Learning 61(1)
- Webb (2007), The effects of repetition on vocabulary knowledge, Applied Linguistics 28(1)
- Chang and Millett (2014), The effect of extensive listening on developing L2 listening fluency: some hard evidence, ELT Journal 68(1)
- Zhang et al. (2022), WenetSpeech: a 10000+ hours multi-domain Mandarin corpus for speech recognition, ICASSP 2022 (transcripts as corrected by Paraformer)
- HSK: the official site of the Chinese Proficiency Test