Details on the HSK7-9 exam are out

I've just registered for HSK 7-9 on 9th May 2026!

 

My Chinese learning has been slow recently, so this will give me a real kick to get into gear. Very excited!!!

On 11/21/2025 at 1:19 AM, Pomelofruit said:

I've just registered for HSK 7-9 on 9th May 2026!

 

Good luck!  Still 5 and a bit months to go so that should be ample time to prepare (the HSK7-9 exams are really quite sparse, both with respect to time and geographical location).  I've been mulling over taking the same exam.

 

Keep us updated on your progress please.  In particular, I'm interested in hearing about study methods, and anything someone who is super-familiar with the HSK6 (2.0 standards) might want to know.

I found two HSK7-9 exam books, but they're in Korean:

 

Here:

 

Quote

 

Here (see also YouTube video）:

 

Quote

 

There's an old HSK7-9 mock exam on YouTube, and my understanding is that it was intended to help students prepare for the first ever HSK7-9 exam.  It contains a restricted collection of questions (and some of them seem unlike the current HSK7-9 exam).  I wanted to make this easier to study than flicking through a YouTube video, so here's a transcript (please forgive both speech-to-text and OCR errors):

 

听力 (listening):
 

Quote

[audio transcription] 价格贵等于东西好吗？朋友的珠宝店开在高档酒店里，最近进了一批绿宝石首饰，虽然正值旅游高峰期，店里每天都挤满顾客，可就是卖不出去，朋友尝试了若干种销售技巧，仍然没效果。最后，他出差前无奈给售货员留了个字条，上面写着，明天把所有绿宝石按原价二分之一出售。几天后，他回来了，发现所有绿宝石都卖完了。起初他并不吃惊，但很快发现售货员竟然粗心地把二分之一误当成了二，所有首饰都按原来价格的两倍卖出去了，这下子他真是百思不得其解了。顾客们的行为不难理解，人们都在一分钱一分货的教育下成长的，这种经验在生活中也得到过多次验证，时间久了，人们不自觉地在心里面把价格贵和东西好画上等号。因此想买好首饰的顾客一看到绿宝石的价格那么昂贵，就觉得它们值得拥有的，所以把它们一抢而空了。其实，上述案例中的心理现象很普遍的。很多时候，人们因没足够时间、精力和能力分辨事物，往往就选择走一条捷径，利用自己的首选经验。殊不知，这样的心理很容易造成判断失误。

 

1. 那家珠宝店每天顾客寥寥无几。 [T/F]
2. 那些绿宝石首饰是按照原价的两倍价格卖出去的。 [T/F]
3. 店主起初看到绿宝石全卖光了感到不可思议。 [T/F]
4. 说话人认为“一分钱一分货”的理念在生活中得到了多次验证。 [T/F]
5. 这段话告诉我们购买贵重饰品时要选择信誉有保证的大型商场。 [T/F]

 

These are "true/false" questions, consistent with the HSK7-9 exam, listening part 1.  The audio here has 385 characters, and the recording is from 1:32---3:07 (95 seconds), which equates to a speaking speed of 243 characters per minute.

 

Quote

[audio transcription]

女：各位观众好，我们今天邀请到的是拥有犬类摄影硕士学位的摄影师张天航，他曾为超过4000只宠物拍摄过写真。天航你好，大家都好奇，你是如何与宠物摄影结缘的。

男：在我眼里，每种动物都是不可思议的存在，我很喜欢观察它们，揣测他们流露的感情。我想通过摄影探索它们的内心世界。

女：想成为宠物摄影师，除了配备专业摄影器材，你觉得还应该具备哪些基本技能？

男：首先要有与宠物沟通的能力。另外，需要有极强的耐心和敏锐的观察力，拍宠物可千万急不得，很多时候需要等待宠物自己进入状态，但适合拍摄的宠物兴奋期有限，无止境等待可能会错过拍摄时机，所以耐心等待和敏捷抓拍共同决定了拍摄效果。

女：据我所知，你一直倡导将宠物作为充满智慧的主角来拍，这在拍摄中是如何体现的？

男：有的人认为宠物摄影就是让宠物做某些动作，然后有摄影师负责按快门就行了，我觉得这是个误区。宠物有思想和智商的，例如边牧犬被认为有相当于7岁儿童的智商，它们会对很多事物做出反应，而我拍摄时更多的是去观察和记录这种反应。

女：在拍摄过程中，你如何与宠物进行沟通，有没有特别的办法和技巧呢？

男：我认为投其所好是拉近距离、消除隔阂的有效方法。可以通过跟主人聊天了解宠物喜欢什么，拿他喜欢的东西来贿赂他，可能是某个物品或某件事。比如有的宠物喜欢被挠肚皮，只要温柔地给它挠肚皮，它就会对你产生好感。

女：你常用的摄影器材有哪些呢？

男：我会根据拍摄对象和场合选择相应的设备。比如拍摄活跃的宠物，选连拍速度快的设备进行动态抓拍。在暗光环境下，选高感性能最佳的设备。使用不同器材拍摄会为我带来更多灵感。比如用小型运动相机时，我会尝试把相机放在水里、雪坑里、石头缝里，去寻找平时无法拍到的角度。

 

6. 男的为什么走上了宠物摄影之路？
A 这个行业的前景很好
B 给宠物拍照容易盈利
C 对宠物的内心很感兴趣
D 不擅长拍摄人物和风景

 

7. 男的认为要想达到好的拍摄效果宠物摄影师需要具备什么？
A 耐心及敏锐的观察力
B 非常高档的摄影器材
C 与其他人沟通的能力
D 勤奋好学的工作态度

 

8. 在男的看来，宠物有什么样的特点？

A 是有思想和智慧的
B 智商完全比不上婴儿
C 很遵守摄影师的指令
D 不受周围的事物影响

 

9. 男的认为怎样才能投其所好，同宠物建立信任呢？
A 饲养各种类型的宠物
B 了解不同宠物的喜好
C 对宠物多多加以称赞
D 得到宠物主人的认可

 

10. 男的是依据______来选择不同摄影器材的。

 

The audio here has 724 characters, and the recording is from 4:37---7:09 (152 seconds), which equates to a speaking speed of 285 characters per minute.  It's similar to the HSK7-9 listening part 2, but there are more "subjective" questions on the HSK7-9 exam.  (Actually, it looks a lot like a HSK6 (listening section part 2) exam question, with the 10th question modified.)

 

Quote

[audio transcription] 中国女排是世界体坛不断创造奇迹的劲旅，其铸造的女排精神是一笔宝贵的精神财富。从20世纪80年代的五连冠，到里约奥运会再度夺取冠军，中国女排拥有过风光和荣耀，也经历过长时间低潮。女排姑娘们却坚持刻苦训练，直到里约奥运会重回世界巅峰，他们用行动诠释和丰富了坚持拼搏的女排精神。多年来，尽管女排队员换了一批又一批，但女排精神一直激励着这支队伍成长、成熟，向新的胜利目标前进，每位世界冠军都要经历从弱到强的发展过程，中国女排也不例外。20世纪60年代，中国乒乓球运动已取得骄人成绩，但排球运动没多大起色，于是中国女排引入了魔鬼式训练法：重视训练难度和质量。比如防守训练，如果一个动作不规范，教练会立刻要求重练几十甚至上百次，从不手软，训练强度和难度远远超过比赛。到1981年，中国女排终于夺冠了。从1981到1985年，中国女排在最高级排球大赛上创下世界排球史上的第一个五连冠。然而，世界体坛竞争激烈，没有永远的冠军，欧美球队的技术和打法在不断升级，中国女排技术更新步伐缓慢，老队员纷纷退役，中国女排陷入了低谷。从1986年到2003年，中国女排有17年与世界冠军无缘。不过这一切磨练了中国女排的意志，使她们变得更加成熟，他们没有被挫折吓倒，而是扎实拼搏和奋斗中不断积蓄着取胜的实力。终于，在2015年世界杯和2016年奥运会上，他们用冠军证明了努力的价值，也证明了女排精神的生命力。

 

11. 女排精神的内涵是______？

 

12. 这段话中提到的魔鬼式训练法指的是什么样的训练？
A 古怪的
B 精简的
C 神奇的
D 艰苦的

 

13. 中国女排第一次获得世界冠军是哪一年？
A 1960 年
B 1981 年
C 1987 年
D 2016 年

 

14. 下列哪项不是女排陷入低谷的原因？
A 队员的更新换代
B 训练经费的短缺
C 其他队伍的不断进步
D 自身技术提升比较缓慢

 

15. 关于中国女排，可以知道什么？
A 将技术创新摆在首位
B 只获得过五次世界冠军
C 十七年后重新回到巅峰
D 不断派队员到欧美球队学习

 

The audio here has 602 characters, and the recording is from 9:02---11:17 (135 seconds), which equates to a speaking speed of 267 characters per minute.  Note the 不是 question (no. 14).

 

阅读 (reading):
 

Quote

鄂伦春族世居中国东北部地区。他们生活的区域山高、林密、溪多，交通不便。到了______六个月的冰雪期，更是寸步难行。但自古以来，勤劳勇敢的鄂伦春人就靠着驯鹿、马、雪橇、滑雪板、桦皮船等传统的交通运输工具，穿林海、过雪原，生活了一代又一代。

 

驯鹿行走平稳，善于穿越森林和沼泽地，曾经是鄂伦春人重要的交通运输工具。无论是迁徙还是打猎，驯鹿都被用来运输物品。但由于速度慢、运载量小，驯鹿逐渐被速度快、运载能力强的马所取代，马成为鄂伦春人生产、生活中最重要的交通运输工具。

 

雪橇和滑雪板在鄂伦春人的生活中也发挥着十分重要的作用。雪橇早期是用狗来牵引的，后来才改用驯鹿和马匹。雪橇在冰雪上滑行，阻力小，滑得快，是一种理想的交通工具。滑雪板在鄂伦春族地区使用的历史悠久，是打猎必不可少的工具，过去几乎每个青年猎手都有一副。每年大雪封闭道路以后，它就派上了用场。滑雪板一般用有弹性的木板制成，中间有一个放脚的皮套。系好滑雪板后，两手各撑一竿，手握撑杆向后推，就会行走如飞。鄂伦春人的滑雪板有长短两种，长的约两米左右，短的一米左右。长的速度快，适宜在又浅又软的雪地上行走。短的灵活，适宜在山上、林中和较硬的雪地上使用。

 

鄂伦春人的水上交通工具是他们自制的桦树皮船。这种船船身狭窄，一般宽不到 1 米，长约 5 米，可乘坐两三个人。划船的时候噪音极小，划着它出去打猎，有利于接近并捕捉猎物。

 

16 根据上下文，第一段的空白处最适合填入的词语是：
A 历来
B 长征
C 长达
D 历经

 

17 驯鹿逐渐被马代替的原因是：
A 马的速度更快
B 马行走更平稳
C 驯鹿不适应沼泽地
D 驯鹿没有运载能力

 

18 下列哪一项不能给雪橇提供动力？
A 马匹
B 人
C 驯鹿
D 狗

 

20 关于桦树皮船，说法正确的是：
A 要从船厂采购
B 用来运载货物
C 划船时很安静
D 空间非常宽敞

 

21 上文主要介绍了鄂伦春族的：
A 交通运输方式
B 发展现状
C 历史沿革
D 社会生活习俗

 

The main text is 589 characters long.  It's like the HSK7-9 exam reading part 1, but the actual HSK7-9 exam reading materials seem more like 1000 characters long, and there are 7 questions per text (not 5).

 

Quote

A 近日，多个国际团队合作的最新研究成果在《科学》中发表。他们利用一系列创新方法，准确测量了人类从出生到老年代谢率的高峰和低谷，首次揭示出了全生命周期的代谢规律。这项研究表明，人的代谢规律呈现出 “升髙 — 下降 — 稳定 — 缓慢下降” 的模式，颠覆了以往人们对 “青春期代谢率高，中年之后代谢下降” 的认知。

 

B 代谢率在经过婴儿初期的激增之后，会以每年约 3% 的速度减缓至 20 多岁，随后便稳定下来，进入正常状态。“尽管青少年时期是生长突增的时期，但数据表明，青少年时期的每天能量需求似乎并没有任何增加。我们本来认为青春期的代谢率会有所不同，但是事实并非如此。” 庞泽说。

 

C 那么真正的分水岭在哪里？研究指出，人体的代谢率在 60 岁之后才真正开始缓慢下降，虽然每年以平均 0.7% 的速度缓慢下降，但到 90 多岁时，人每天需要消耗的卡路里里仍有中年时期的 74%。

 

D 研究还发现，20 多岁至 50 多岁这一时期的能量消耗是最稳定的。即使在怀孕期间，伴随着婴儿的成长，孕期女性的能量需求仅仅是随着体重的增加而增加，其自身的能量需求并没有出现大幅上升。“也许有人告诉你，30 岁以后身体的各项机能都会下降从而导致体重的增长，但作为家庭和社会中流砥柱的中年人，发福的因素有很多，我们的研究表明，最起码代谢率变化并不是原因之一。” 斯彼克曼表示。

 

E 许多人都有类似的感受：一段时间经常熬夜或作息不正常，就感觉老得特别快。有实验显示，缺乏睡眠的受试者体内有 711 种基因的功能发生了改变，其中就涉及到新陈代谢的相关基因。专家称，睡眠不足可能降低人体对葡萄糖的调节能力，同时减少瘦素分泌，增加饥饿素分泌，使人容易感到饥饿，间接造成脂肪代谢异常，建议每天至少要保证 7 小时的睡眠。

 

F 研究数据表明，婴儿的代谢率是最高的。婴儿在出生后的前 12 个月里，对能量的需求会迅速增加，因此到一岁时，婴儿单位体重消耗的卡路里可比成人的代谢率高 50%。专家说 “部分原因可能是在他们生命最初的几周基本上是久卧不动的，到一岁左右才开始真正活跃起来。与此同时，婴儿的静息代谢率也大幅度上升，这可能与他们惊人的生长率有关。” 婴儿期如此高的能量需求，可能导致婴儿期的成活率十分依赖能量供给，如果在婴儿期这个关键的发育时间窗口能量供给不足，就很难成活，即使成活下来也很难正常生长到成年。

 

G 虽然能量需求随时间变化的机制目前尚不完全清楚，但该研究改变了以往人们对代谢的认知，表明我们可能需要重新审视发育阶段和衰老阶段的能量需求和营养策略。

 

A -> [...] -> [...] -> [...] -> [...] -> [...]

 

This is a new feature of the HSK7-9 exam, where the student is given 7 paragraphs and arranges 6 of them in order, with 1 paragraph already given (in this case A), and we throw one away.  The text is 1074 characters long.  It seems very much like the HSK7-9 exam, reading part 2.

 

Quote

20 年前我去山西平遥参观了中国票号博物馆，我在那次参观中受到很大启发，简单地说一下。

 

当年票号（就是银行总部）在平遥，东家（就是股东）办了这个银行，办银行需要在全国各地办分号，也就是现在的银行分行，那么他就需要选择一批德才兼备的掌柜（就是我们现在说的职业经理人）到各地去经营。

 

东家选择了他认为合适的掌柜，派到外地去经营。在当时的条件下，是无法监管的，因为交通极其不方便，更谈不上通讯设施。

 

那么在那种情况下，委托代理的矛盾就非常尖锐，就是说：我选择了我认为很可靠的掌柜，但他是不是忠诚于股东？是不是勤勉？是不是廉洁？我无法监管；一旦我看错了人，那我这个分号可能就会出现重大的经营问题甚至破产倒闭。这个委托代理矛盾比我们现在遇到的委托代理矛盾要突出得多、严重得多。我们现在都有这么多的监管措施了，比如公司治理、审计监督、法律、纪检巡视等，仍然没有很好地解决这个问题。

 

可是 300 年前我们的老祖宗，他们创造了一个什么制度来解决这个问题呢？“身股银股” 制度。就是说，东家出资人跟掌柜说：“你到上海去，你好好经营，如果有了利润，六四分成。” 就是掌柜和伙计们分 60% 的利润，东家分 40% 的利润。

 

这是什么概念？就是干活儿的人拿大头，出资人拿小头。这个制度起什么作用呢？就使得我们讲的委托代理矛盾当中信息不对称、利益不一致的问题从根本上解决了：因为掌柜和伙计们不再是为东家干活儿，而是给自己干活儿，所以他们的利益跟东家的利益是一致的。

而且一定意义上讲，是东家搭了掌柜和伙计的便车：你们好好干活儿，最后有利润我分 40%；有亏损、有风险谁承担？东家承担。比如说被抢了、被火烧了等意外的损失，掌柜跟伙计是不承担责任的，按照史料上讲，是负盈不负亏。

 

那么结果是什么情况？据史料记载，当年我们国家的票号发展得非常好，称得上 “富可敌国”。当时清政府没钱时，就去向票号借钱，一借就是几百万两白银或者上千万两白银，足以说明它已经到了富可敌国的程度。

 

也就是说，当年我们票号的利润分配制度是很好的、发展得是很好的。所以，这也证明了这个制度是能解决委托代理当中的问题的，使得出资人的利益和干活人的利益形成利益共同体，大家一起把企业做好，共同创造利润、共同分配利润。他们当时分配利润是每三年分一次，每一股分的利润非常丰厚。

 

27 票号分号在现在指的是什么？ [...]
28 为什么当时监管外地的掌柜比较困难？  [...]
29 画线词语 “这个问题” 指的是什么问题？   [...]
30 按照 “身股银股” 制度，企业出现风险由谁承担后果？   [...]
31 画线词语 “拿大头” 是什么意思？   [...]
32 请用一个成语回答当年的票号的发展程度。   [...]

 

 

This (non-multiple-choice) question is intended to be answered by writing your own answer.  The text is 965 characters long.  It's like the HSK7-9 exam, reading part 3, but the actual exam has 7 questions per text (not 6).

 

写作 (writing):
 

Quote

33 请对图表进行描述与分析，写一篇 200 字左右的文章，限时 15 分钟。

 

公众环保意识形成的影响因素

对气候和环境变化的自身感受：64.5%
环境生态相关的纪录片：59.0%
环境生态相关的新闻媒体报道：52.5%
政府和环保机构的公益宣传：47.5%
相关环保公益活动的亲身体验：36.3%
综艺娱乐节目传达的环保理念：26.3%

 

This another new feature of the HSK7-9 exam.  They give you a plot/diagram/table or something, and you are to describe it in around 200 characters.  This looks consistent with the actual HSK7-9 exam, writing part 1.

 

Quote

34 话题作文，限时 40 分钟

 

诸葛亮的《诫子书》是中国古代家训中的名篇，“静以修身，俭以养德。非淡泊无以明志，非宁静无以致远。” 意思就是 “以宁静来提高自身的修养，以节俭来培养自己的品德。不恬静寡欲无法明确志向，不排除外来干扰无法达到远大目标。” 请写一篇 600 字左右的文章，谈谈你对节俭的认识并论证你的观点。

 

Another new feature of the HSK7-9 exam.  Judging from the examples I've seen, they give you a quote from classical literature, and a translation of that quote in modern Chinese, then ask you to write an argumentative piece of length about 600 characters.  This looks consistent with the actual HSK7-9 exam, writing part 2, which requires students to:

 

Quote

express and defend opinions on given topics

 

This is kind of similar to some Gaokao 议论文 questions I've seen, except on the Gaokao, they don't translate it into modern Chinese, and Gaokao students need to write at least 800 characters (they can go over).

 

翻译 (translation):

 

Quote

35 请将下列材料译写成中文，限时15分钟。

 

Whether it’s a tricky maths problem or an unexpected bill, daily life is full of stressful experiences. Now researchers have found that humans produce a different odour when under pressure - and dogs can sniff it out.

 

While previous studies have suggested canines might pick up on human emotions, possibly through smell, questions remained over whether they could detect stress and if this could be done through scent. “This study has definitively proven that people, when they have a stress response, their odour profile changes,” said Wilson, a PhD student, and first author of the research.

 

Wilson added the findings could prove useful when training service dogs, such as those that support people with post-traumatic stress disorder.

 

The HSK7-9 exam introduces translation: for the HSK7-9, the translation is always from English (or some other language) into Chinese.  This mock exercise seems consistent with the HSK7-9 exam translation section 1 (written translation).

 

The second half continues in a separate YouTube video:

 

Quote

36 Everyone has bad days at work. You hit an unexpected complication with a project you’ve been working on for weeks. A colleague snaps at you in a meeting. You forget to put something on your calendar and leave a key client waiting.

 

Whatever the root of your workplace problem, one effect it has is to make you feel more alone. In those bad moments, you feel as though your colleagues are upset and you can only rely on yourself.

 

The task is to translate this to Chinese orally.  The audio says there is 1 minute of reading time, and 2 minutes of translation time.  This corresponds to the HSK7-9 translation section, part 2.

Quote

37 一、预约时段

选择参观日期

2022.02

一（MON）	二（TUE）	三（WED）	四（THU）	五（FRI）	六（SAT）	日（SUN）
14   闭馆	15   有票	16   有票	17   闭馆	18   闭馆	19   闭馆	20   闭馆
21   闭馆	22   有票	23   不可预约	24   不可预约	25   不可预约	26   不可预约	27   不可预约

 

入馆时段

    09:00-13:00（已满）
    13:00-16:00（有票）

 

二、参观注意事项
（一）仅限个人预约，无预约不得入馆参观。首都博物馆微信公众号是个人预约参观的唯一平台，每位观众须进行实名预约，免费申领门票。
（二）个人预约观众从北门入馆，凭本人二代身份证（外籍人员凭护照）验证预约信息并领取门票。

（三）入馆时须完成：体温检测、“北京市（健康宝）” 健康码检查；外籍人员出示 “Health Kit（健康宝）” 检查。

（四）体温大于等于 37.3℃、健康码显示异常、未预约这三种情况中出现任何一种的观众，不得入馆参观。

（五）观众进入馆区后，须全程佩戴口罩，未戴口罩者谢绝参观。

 

你的外籍朋友爱莎 2 月 14 号到 27 号来北京旅游，你帮她预约了一张 2 月 15 号下午的参观票，但那天你有事不能陪她去，请你告诉她预约信息以及参观的注意事项。

-----

 

Actually, while I'm here, there's also an official example HSK7-9 exam (also incomplete).  The pdf is here and the audio for the listening section is here.  The oral section also has an audio component, but I couldn't find it.  I'll include the transcript for the listening section (as it's on the pdf for the other sections):

 

听力 (listening): 

 

Quote

[audio transcription] 近年来，通过科技创新缩减地区数字鸿沟，推动可持续发展目标成为各界关注的话题。全国多地利用5G等科技发展智慧旅游，为乡村振兴注入强大动力，贵州省苗族侗族自治州雷山县西江千户苗寨景区内，游客戴上混合现实眼镜解谜寻宝，探索5G加智慧苗寨的全新体验，特色街区实景与动感数字画面结合，让苗文化感受更深刻。游客王女士说，景区工作人员介绍，景区智慧旅游信息化建设项目依托5G网络，实现信息服务集成，人们能方便观看景区实时地图和路线导览，查询智慧停车信息、今年春节期间，西江千户苗寨接待游客14.2万人次，同比增长482%，实现旅游综合收入1.4亿元，同比增长532%。同时，苗寨举办了全国村晚示范展示活动，通过5G网络云推介当地乡村村落民俗风情和丰富旅游资源，吸引超过千万人次观看。随着5G通信基础设施覆盖扩大，5G与人工智能、虚拟现实、大数据、边缘计算等技术应用深度融合，让乡村插上了智慧转型的翅膀。

 

1. 智慧旅游的兴起，有助于全民科学素养的提升。 [T/F]
2. 西江千户苗寨景区有多个历史谜团尚未解开。 [T/F]
3. 今年春节期间，西江千户苗寨景区累计接待游客数同比增长 482%。[T/F]
4. 5G 与虚拟现实、增强现实等技术和应用的深度融合，使得 5G 等通信基础设施覆盖范围不断扩大。[T/F]
5. 这则新闻主要谈的是 5G 网络与旅游的结合助力了乡村振兴。 [T/F]

 

The audio here has 395 characters, and the recording is from 1:14--2:48 (94 seconds), which equates to a speaking speed of 252 characters per minute.

 

Quote

[audio transcription]

女：徐教授您好，我们还是就看刚刚看到的短片来聊，请问电子嗅觉的工作原理是什么？

男：电子嗅觉本身是一个学科交叉的产物，因为电子嗅觉其实是模拟人的嗅觉来工作的。我们知道人类有嗅觉是因为有嗅觉受体，同时还有很多跟嗅觉相关的神经细胞等等。其实我们嗅觉所感触到的是气味，气味就是气加味。所谓气就是空中漂移的物质分子，味是我们的嗅觉受体感知。人工智能算法能知道我们感知到什么味道，同时还能提高感知的效率、速度、敏感度等等。

女：这个机器是如何完成人类嗅觉神经工作的？

男：嗅觉传感器上有十几个不同传感单元，每个里面都放了感应材料。当气体，也就是所谓的气味，沿着小孔进去后，会和涂在上面的感应材料发生反应，引起一些物理参数的变化，这些变化能让我们找到它和数据库里味道数据之间的差异。我们做的是模式识别和模式匹配。如果匹配上我们就会知道，大概是什么物质或者什么味道。

女：我们都知道对于气体成分的识别和监测已经有非常完善的设备，像很多家庭都安装了燃气报警器，那这样的话，电子嗅觉的优势在哪里呢？

男：实际上烟雾报警器也好、专用报警器也好、还有家用二氧化碳或煤气报警器，它们都是嗅觉传感器，只不过以往大家更多做的是单一气体的，我们现在做的是相对通用的，也就是说，它可以闻到很多种味道。比如刚才视频里的机器狗，它可以准确闻出10种味道，但在视频演示中只测试了10种，实际上能闻出更多。

女：我们刚才说的都是电子设备的嗅觉，我比较好奇的是它能不能精准模拟出一款气味的味道，比如当我在网上购买香水这样的产品时，只能通过文字描述来简单地了解这个味道是什么样子的。未来会不会有这样一个终端，当我想选购这款香水的时候，它可以模拟出相应的味道来让我闻。

男：从理论上讲，它可能是的，就像您说的，我们可以通过某种方式来合成新的味道，或者是合成已有味道。比如我们曾探索可否设计和制作一款气味照相机，目的是希望我在拍一个物体的时候，不仅能拍到它的图片，还能拍到它的味道。比如说，拍一个苹果，我拍下以后，可以把苹果的味道储存在胶片上，然后将来，看这个照片的时候，就可以闻到它的味道。当然，现在这只是个概念而已，但我相信随着时间推移、技术的进步，这件事情在不久的将来应该是能够实现的。

 

11. 电子嗅觉实质上是通过______来完成工作的。

 

12. 嗅觉传感器是如何识别气味的？
A 放置在感应材料中进行比较
B 识别气体成分进行数据库匹配
C 凭借嗅觉系统和神经细胞来区分
D 参照气味引起的物理变化和化学反应

 

13. 与烟雾报警器相比，电子嗅觉有什么优点：
A 可辨别多种气味
B 能实现远程监控
C 具有自动报警功能
D 可检测有毒有害气体

 

14. 气味照相机把气味存储在______上。

 

15. 关于模拟味道的终端产品，男的是什么看法：
A 从理论上来讲是不可行的
B 倘若被滥用会带来巨大风险
C 目前未落地，但未来有望实现
D 只是炒作罢了，根本无法实现批量生产

 

16. 根据对话，下列哪项正确？
A 机器狗顶多能闻出 10 种气味
B 人工智能算法能够增强感知灵敏度
C 电子传感器可以通过文字描述合成香水气味
D 二氧化碳报警器属于家庭普遍安装的专用报警器

 

The audio here has 929 characters, and the recording is from 4:22--7:22 (180 seconds), which equates to a speaking speed of 309 characters per minute.

 

Quote

[audio transcription] 赏花作为文化活动，在漫长岁月中并非一成不变，而是随着中华文明整体进步，不断变换的风貌，内涵逐渐丰富。中国自古农业发达，古人在长期发展农事过程中，对植物培养有浓厚兴趣和细腻感情，特别欣赏花木的美。中国自然条件得天独厚，大部分地区四季分明，经历三四个月草木萧瑟后，春天回归显得美好，山水秀美，草木茂盛，处处都是美景，从先秦时代起，人们在春天自发到山间水畔赏花的风俗，人们赏花以自然生长的花木为对象，会集体到郊外赏花，在山水花木间尽情游玩，赏花是盛大的郊游活动，不论男女老少，都一起到山水间游玩，特别是先秦时期，青年男女会一面采花，一面寻觅心上人，彼此中意，就会互赠亲手采摘的香花美草，爱情在春花的见证下绽放了。到了盛唐，官方机构或民间力量在城内外风景优美的地方栽培主题性花木，加建亭台楼阁，形成半人工化的风景区，供全民在春天一起赏花。曲江是当时著名的赏花风景区。除了赏花，人们还会在花下野餐、唱歌、奏乐，进行跑马、放风筝等各种游戏，风景区商贩云集，曲艺人可随处露天表演，给游人增添乐趣，赏花已然成为中国人别样的狂欢节。从唐后期开始，随着私人园林的成熟，又出现了一项极具公德心的习俗，春天花开时，所有私人园林都要向公众开放，大众自由来去，享受园林之趣。宋代城市日益发达，汴京、临安等大城市里私人园林众多，春天，到各家园林看花成了城市生活的重要内容，人们不仅到郊外欣赏春光，还会把城里城外的著名园林一家一家都逛遍，游园看花过程中优美的园林景致给予人们艺术陶冶、造景等的美学魅力，让人们的艺术修养在潜移默化中得到提高。到了明清，私家园林的主人普遍会在园内轩堂中陈列自家收藏的文物、古玩、书法、碑帖、古画等，相当于举办小型文化展览会，供游人欣赏，游园看花变成了文化交流活动，有心人借着看花欣赏艺术品，增长见识，提高品位。私家园林如此费心费力，却纯属公益性质，主人不会收费，社会上形成了一种富有人情味的风气，游园的人会给园丁一点小费，叫扫花钱，意思是感谢园丁一年到头侍弄花木，让大家能够尽享春光。

 

23. 下列哪项不是中国古人赏花风俗产生的原因？
A 气候四季分明
B 自然环境优越
C 对华丽生活的向往
D 对植物培养情有独钟

 

24. 先秦时期，年轻人是如何在赏花时表达喜爱之情的？
A 邀约一同赏花
B 采花赠给意中人
C 放飞风筝寄托思念
D 借助乐曲传递情谊

 

25. 以下那种赏花场景最可能出现在盛唐时期？
A 在曲江赏花听曲
B 在亭台举行祭祀仪式
C 在私家园林欣赏古玩
D 风景区内商贩寥寥无几

 

26. 游园看花过程中，普通人的什么获得了提高？

 

在宋代，游园看花过程中，人们的______得到了提高。

 

27. “扫花钱”是什么钱？
A 园林的维修费
B 感谢园丁的小费
C 主人支付给园丁的报酬
D 实质上是进入园林的入场券

 

28. 关于赏花的变化趋势，下列哪项正确？
A 以民间自发为主到以官方主导为主
B 赏花场所从人工景区转移到近郊野外
C 赏花群体从知识分子发展到平民百姓
D 赏花活动从赏花游乐到兼顾文化交流

 

The audio here has 842 characters, and the recording is from 9.32--13:09 (217 seconds), which equates to a speaking speed of 232 characters per minute.  Note the 不是 question Q.23, and how the audio and text in the pdf are not identical for Q.26.

Hmm... this HSK7-9 mock exam collection suddenly showed up on Jingdong:

 

 

This is 《HSK7-9级全真模拟测试题集》， and despite the near-identical name, I don't think it's the same as 《HSK7-9级全真模拟题集》.

 

And there appears to be a HSK7-9 graded reader:

 

 

It's called 《最炫中国风》.  I'm not sure what the purpose of a graded reader is at the HSK7-9 level (we can just read anything we want (?)).

I found 《最炫中国风》on Taobao and read the table of contents there. It looks like the typical Chinese reader filled with short articles, stories, and excerpts.  It looks lame. 
 

I passed HSK 6 a few years ago. I had probably read 20+ novels or narrative non-fiction books at that stage, and like most people, I found the reading part of the test and in particular the need to read at a fast enough speed pretty challenging. The only way to get over that is through high-volume extensive reading. I doubt a reader filled with sliced-and-diced short readings would be helpful preparation for HSK 7-9. 

On 1/1/2026 at 3:07 PM, becky82 said:

This is 《HSK7-9级全真模拟测试题集》， and despite the near-identical name, I don't think it's the same as 《HSK7-9级全真模拟题集》.

both 模拟考试, that first part of book, first part 5 exams.

The new HSK 3.1 textbook for HSK 1 is out now (price point 75 RMB) and the ones for HSK2 and 3 will be available in March 2026.

 

There is a slide preview of HSK 1 book available, if you click on the above link. In the HSK 1 book there seems to be an "AI assistant" guiding learners, but all content pages of the book related to that are skipped in the preview.

 

Each book comes with additional digital content (audios, videos, digital textbook, standardised slides to be used by teachers). In the video under the link it looks like the new digital platform will also feature AI to grade speaking abilities. The first batch of the digital content will be released in February.

 

Additionally, each book comes with a practice book that includes regular practice and mock exam sections.

 

 

 

HSK levels 1-3 will consist of one volume each, HSK 4-6 will have two volumes each, and HSK 7-9 is split into 5 different textbook covering different skills.

 

I’m currently using these two mock paper compilations to prep for my HSK exam in May. They have different content.

 

I’m finding the speaking part the hardest. The model answers aren’t particularly helpful to me - it’s basically written Chinese rather than spoken Chinese, so I’ve been asking Deepseek to generate answers for me based on the spoken Chinese level of a middle schooler. 

 

Will need to do a lot more prep!

 

 

@Pomelofruit, did you study all the HSK7-9 vocab?  If so, any advice?

 

I'm currently going through it now (5600 words, 2025 3.0 standards) and it's a total grind.  At my peak, I can maybe process 200 words per day (study them to a HSK7-9-suitable depth), and I can only achieve this rate because ~50% of them I already know.

On 2/28/2026 at 8:39 AM, becky82 said:

did you study all the HSK7-9 vocab?  If so, any advice?

I’ve only taken HSK6 in the previous iteration of the exam, but I nevertheless think that when one gets to that level or higher, direct cramming of vocabulary lists is counterproductive. I think that at least in theory, there are more productive ways of preparing for HSK, and of using the word lists. I outline my reasoning below. 

 

1. The consensus of HSK6 takers is that reading speed is crucial. Many (most?) people’s lowest score is on the reading section, and a commonly cited reason for this is simply struggling to read all the passages and questions in the given time.
2. The only reliable way to boost reading speed is high-volume extensive reading of material where one already understands 98 or 99% of the words/characters on the page. That’s the level of lexical knowledge needed to be able to infer the meaning of and thus incidentally learn the unknown 1-2% of words on the page.
3. For any language, cramming of high frequency vocabulary can mutually support learning from extensive reading because the target vocabulary is seen and heard so frequently. But the general academic consensus is that once one is into the mid-frequency vocabulary (i.e., the massive amount of vocabulary that native speakers learn between about age 5 to 18), rote drill will not lead to sufficient retention to break the glass ceiling above which one can comfortably read most adult novels, magazines, newspapers, etc. There is just too much damn vocab for that to work. 
4. For English and most other languages, few published graded readers go beyond high frequency vocab. So extensive reading programs will use the BNC/COCA word family lists (1k up to 24k) and a freeware called Antwordprofiler (similar functionality to Chinese Text Analyzer) to lexically profile books so that a second language reader who has taken a receptive vocab knowledge test can identify which books are appropriate to their level.
5. A few corpus linguists have created 1000 word high and mid-frequency lists for Chinese (I think going up to the 10k most frequently used words), but I’m not sure how valid those lists really are as I have not seen any peer reviewed articles on how they were created. I think the HSK level lists are probably the best lists to use right now if one wants to lexically profile books.
6. I think the best use of the HSK 1-9 lists will be to profile books for extensive reading, and then focus one’s reading on books where one already knows about 98% of words on a typical page, or at least enough to read fluently enough without looking much stuff up. That strategy will kill two birds with one stone: One would improve reading speed and stamina, AND build receptive vocab knowledge faster than from the method of cramming from lists.

 

Has anybody used the HSK 7-9 lists with Chinese Text Analyser? 

 

I’ve never used Antwordprofiler with Chinese word lists and texts, but I recall its creator stating that it is possible

I just noticed this for the HSK7-9:

 

Quote

Report Score: The ability value for each skill of the test takers' Chinese proficiency is calculated by using the IRT model, based on the raw scores of the test takers. Besides, the measurement model is further used to obtain the report score of each skill range from 0 to 100.

 

The IRT model is a statistical model where students have an ability value θ on a standard distribution that represents their actual skill level.  Their ability level is estimated by calculating the most likely θ that would give the student's exam results (incorporating how hard the questions are, and how other students perform), and this is then somehow (it's not explained how) converted to 0 through 100, giving the student's marks.

 

Importantly, your mark is not your raw score: two students can get the same number of answers correct, and get different final marks depending on which questions were harder.

On 2/28/2026 at 8:39 AM, becky82 said:

@Pomelofruit, did you study all the HSK7-9 vocab?  If so, any advice?

 

I'm currently going through it now (5600 words, 2025 3.0 standards) and it's a total grind.  At my peak, I can maybe process 200 words per day (study them to a HSK7-9-suitable depth), and I can only achieve this rate because ~50% of them I already know.

Even though I bought the vocab books, I haven’t opened them once because I realised that studying vocab is not as productive as doing the past exams or general reading of various articles (I paid for TCB The Chairman’s Bao app and select the level 7 articles only, or I browse articles on WeChat).

 

One thing I do on TCB is to speed through an article ignoring all the unknown vocab, answer the questions at the end and then go back to re-learn anything (the app has in-built questions). That way, my reading speed improves and I can use context to infer any unknowns. This has been helpful!

I’m not sure if the mock exams are representative, but I can just about finish the reading section and answer questions to a decent degree. 
 

For me, the hardest part is the spoken translation. The sentences are complex and in written Chinese format rather than spoken language, and the one minute prep is way too short! I can barely read the English text before needing to translate. 
 

Whilst my written Chinese is not great, it’s improved since I’ve been using Deepseek to correct and rewrite my articles. If only Deepseek could improve my real-time translation skills!

Thanks.  It seems people are skipping the vocab study.  The additional 5600 words is a lot, and is questionably beneficial (they can't all appear on the HSK7-9 exam).  However, I personally don't think I will feel prepared for the HSK7-9 exam without studying the vocab in full (for HSK5 and HSK6, learning the vocab was always "step 1").

 

I've been grinding the the vocab recently, working backwards (I've gotten through Z, Y, X, W).  I must admit, it's pretty tedious.  I find I already know 50% well enough, about 30% need debugging or have room for improvement, and 20% are unfamiliar (or somewhere in this ballpark), so there are some gaps in my knowledge.  I'm finding the other drawback of this approach is that it feels like I'm jumping from HSK6 to HSK9 vocab in one step.  And I probably won't be able to make the May exam if I complete the vocab.  The other side of the coin is: there are no more HSK words after this, so it feels "done" (well, if I really wanted to, I could also study the previous 2021-version HSK 3.0 too, and the HSK 1.0 vocab, but you get the idea---it comes with a "finish line").  And I'm aiming to use these skills beyond HSK exams, and I get enough input so that these words (as a whole) are encountered quite regularly.  Still, I don't know if this is the optimal decision.

 

Yeah, translation was a bit of a surprise addition: for years I've been learning Chinese using Chinese, believing that was the correct thing to do, and now translation is required.  It's a task I'm not especially interested in.  And I'm not sure if I'm meant to be literally/directly translating or not.  And are the people who evaluate translations strong in both Chinese and English?

 

I've been using DeepSeek to help with my writing too (it seems to be one of the better choices of AI).  I've been posting my writing on 简书 too.  I'm finding over time I need DeepSeek's help less and less (or maybe I know what it's going to say before it says it).

So you haven’t registered for the May exam yet, @becky82?

 

I agree with your drawback re: jumping from HSK6 to HSK9 with your vocab list, unless you’re aiming to get a 9 first time?

 

For me, I’m expecting an 7 but aiming for an 8 with the view to retake the exam further down the line once my Chinese improves. I find that having an exam gives me the pressure to study, otherwise I’m lazy and get by on the Chinese I know already…

 

RE: your question about direct translation vs “convey the meaning” translation, I have no idea! There’s a difference in the 参考答案 between the two HSK exam books: the green book emphasises direct translation whereas the red book is much, much less direct. 

No I haven't registered.  I'm renewing my passport at the moment, so I'm having to travel.  I could still take the exam when I get back, just for the experience (so I know what to do next time), as one of the exam venues is Weihai, where I live.  But I won't be fully prepared for it (although there is HSKMock).

 

I don't expect a HSK9 the first time; I'd be happy enough with HSK7.  It's kind of weird having HSK6 "demoted" to intermediate level.

 

One thing I learned from the HSK6 is to watch out for "HSK lock in".  If I take the HSK7-9, and don't do especially well, then does that morally commit me to passing it at some point in the future?  And what if the time required for me to pass the HSK7 is longer than I currently envision?  I don't want to get stuck in this quagmire again.