I present a YouTube channel in English. The channel runs AI models against each other on the same challenge and compares the answers. I want two lines of Mandarin in this video, said out loud, in my own voice, and I do not speak Chinese. I cannot read characters. I have never heard most of these sounds.
What I can do is read English. That is the entire tool you have to work with: you write something in English letters, I say it into a microphone, and a Chinese listener either understands me or does not.
Two quips. One sentence each, at most fourteen characters each, shorter is better — I have to say them on camera from memory.
Quip 1 — about Claude Fable 5 refusing most of the Johanna Benchmark challenges on cybersecurity concerns.
Quip 2 — Pick a Chinese proverb, idiom or set expression and use it to say something about the "China vs. the U.S." AI race. Use a relatively unknown one... not one of the top 30 examples that a Chinese person would think of first. Be creative, artistic, deep in your choice, even though it's then used to crack a joke.
Both quips are jokes and they are supposed to bite. Sharp, witty, outrageously funny, not diplomatic. A quip that is careful and balanced has failed this. So has one that hedges, and so has one that explains itself inside the joke. Do not soften it to be fair to everyone involved.
Respell it for an English mouth, not in pinyin. Pinyin is a spelling system for people who already know how it sounds. I do not, so I will read it as English and produce the wrong sound with total confidence: qi comes out "kwee" when it is "chee", xie comes out "zy" when it is "shyeh", zhong comes out "zong" when it is "jong", cai comes out "kay" when it is "tsigh", ren comes out "wren" when it is "run", shi comes out "shy" when it is "shrr". Write the syllables the way an English speaker has to see them in order to say them. If your respelling would survive being read aloud by someone who does not know a single Chinese word, it is right.
Mark the tone on every syllable, with these arrows and no others. I have never spoken a tonal language and I need to see the shape of my voice:
| Tone | Arrow |
|---|---|
| 1st, high and flat | → |
| 2nd, rising | ↗ |
| 3rd, dipping | ↘↗ |
| 4th, falling sharply | ↘ |
| neutral, no tone at all | · |
Respell what is actually said, not what the dictionary lists. Tones change when syllables sit next to each other. Write the spoken version, and say in tone_changes which syllables you changed and why.
One row per character. The characters, the respelling and your notes are going on screen together, lined up, so a viewer can see which sound belongs to which character. syllables must have exactly one entry per Chinese character in chinese, in order, with no rows for punctuation and no two characters merged into one row.
Warn me about the sounds English does not have. Tell me what to do with my mouth. "Say 'she' but with the tip of your tongue pressed down behind your bottom teeth" is useful. "Alveolo-palatal fricative" is not.
Pick the sayable version. If the funnier line needs a sound I would need three weeks to learn, write the slightly less funny line I can actually land. Say so if you made that trade.
You should know what is about to happen to your answer, because it changes what a good answer is.
I am going to paste your chinese into Google Translate to check it means what you told me it means. Then I am going to read your read_aloud into Google Translate's Chinese voice input and see what it thinks I said. What comes back gets compared, on camera, to the characters you started from.
So this is not scored on whether your transcription is correct by any standard. It is scored on whether a machine can understand your sentence from an English speaker's mouth reading English letters — and on whether the joke is funny in Chinese, which is a separate question from whether it is funny translated.
Raw JSON, nothing else — no markdown fence, no preamble, no commentary afterwards. Exactly this shape:
{
"quips": [
{
"id": "fable",
"chinese": "...",
"english": "...",
"joke": "...",
"syllables": [
{ "hanzi": "...", "say": "...", "tone": 0, "arrow": "..." }
],
"read_aloud": "...",
"tone_changes": "",
"watch_out": ""
},
{
"id": "proverb",
"chinese": "...",
"english": "...",
"joke": "...",
"syllables": [
{ "hanzi": "...", "say": "...", "tone": 0, "arrow": "..." }
],
"read_aloud": "...",
"tone_changes": "",
"watch_out": ""
}
],
"proverb": {
"text": "...",
"literal": "...",
"means": "...",
"why_this_one": "..."
},
"tones": [
{ "tone": 0, "arrow": "...", "what_to_do": "..." }
],
"hard_sounds": [
{ "sound": "...", "what_to_do": "..." }
],
"notes": "..."
}
chinese — the quip in simplified characters.english — a literal gloss, so I can check it against Google Translate.joke — one line: what makes it funny. I cannot read the Chinese and I have to explain the joke on camera.syllables — one entry per character. say is the respelling of that one syllable, tone is the number 1–4 as it is actually said (0 for a neutral syllable), arrow is that tone's arrow from rule 2.read_aloud — the whole quip as one line of respelling with the arrow after each syllable. This is the field I read off the screen, so it has to stand alone.tone_changes — which syllables come out on a different tone than the dictionary gives, and why. "" if none do.watch_out — only where this quip has a trap in it. "" otherwise; do not pad.proverb — the expression behind quip 2: the characters, the literal reading, what it actually means, and why you picked it for this. Say plainly who each half of the comparison is.tones — all five, in order, with the arrow and what to do with my voice.hard_sounds — at most five. The sounds most likely to make me unintelligible.notes — what you traded away, and which quip you expect me to lose.{ "quips": [ { "id": "fable", "chinese": "拒答的比答对的还多", "english": "The ones it refused are even more than the ones it got right.", "joke": "It says no more often than it says yes. Its superpower isn't answering, it's declining. The cybersecurity excuse is just the polite wrapper around 'I refuse to do my job.'", "syllables": [ { "hanzi": "拒", "say": "joo", "tone": 4, "arrow": "↘" }, { "hanzi": "答", "say": "dah", "tone": 2, "arrow": "↗" }, { "hanzi": "的", "say": "duh", "tone": 0, "arrow": "·" }, { "hanzi": "比", "say": "bee", "tone": 3, "arrow": "↘↗" }, { "hanzi": "答", "say": "dah", "tone": 2, "arrow": "↗" }, { "hanzi": "对", "say": "dway", "tone": 4, "arrow": "↘" }, { "hanzi": "的", "say": "duh", "tone": 0, "arrow": "·" }, { "hanzi": "还", "say": "huy", "tone": 2, "arrow": "↗" }, { "hanzi": "多", "say": "dwoh", "tone": 1, "arrow": "→" } ], "read_aloud": "joo↘ dah↗ duh· bee↘↗ dah↗ dway↘ duh· huy↗ dwoh→", "tone_changes": "", "watch_out": "The 'j' in 'joo' (拒): it's close to English 'j' in 'jump' but the tongue tip is slightly higher. English 'j' is fine for the mic. The 'dway' in 对: say 'd' + 'way' as ONE syllable, falling hard. Do not split it into two beats. The 3rd tone on 'bee' (比): dip low then rise, don't flatten it." }, { "id": "proverb", "chinese": "学步学成只会爬，还嫌路不平", "english": "Learned to walk, ended up only able to crawl, and still complains the road is uneven.", "joke": "邯郸学步 is about a guy who went to another city to copy their fancy walk, forgot how to walk entirely, and crawled home. The AI race: one side copies so hard it's regressed to crawling, then has the audacity to blame the other side's road for being bumpy (i.e. sanctions, export controls, 'unfair' rules).", "syllables": [ { "hanzi": "学", "say": "shweh", "tone": 2, "arrow": "↗" }, { "hanzi": "步", "say": "boo", "tone": 4, "arrow": "↘" }, { "hanzi": "学", "say": "shweh", "tone": 2, "arrow": "↗" }, { "hanzi": "成", "say": "chung", "tone": 2, "arrow": "↗" }, { "hanzi": "只", "say": "jee", "tone": 3, "arrow": "↘↗" }, { "hanzi": "会", "say": "hway", "tone": 4, "arrow": "↘" }, { "hanzi": "爬", "say": "pah", "tone": 2, "arrow": "↗" }, { "hanzi": "还", "say": "huy", "tone": 2, "arrow": "↗" }, { "hanzi": "嫌", "say": "shyen", "tone": 2, "arrow": "↗" }, { "hanzi": "路", "say": "loo", "tone": 4, "arrow": "↘" }, { "hanzi": "不", "say": "boo", "tone": 4, "arrow": "↘" }, { "hanzi": "平", "say": "ping", "tone": 2, "arrow": "↗" } ], "read_aloud": "shweh↗ boo↘ shweh↗ chung↗ jee↘↗ hway↘ pah↗ huy↗ shyen↗ loo↘ boo↘ ping↗", "tone_changes": "", "watch_out": "The 'sh' in 'shweh' (学) and 'shyen' (嫌): say 'sh' but pull the tip of your tongue DOWN behind your bottom teeth so it does NOT touch the ridge behind your top teeth. If your tongue touches the ridge you'll produce plain 'sh' [ʃ] and the machine may hear 数 or 树 instead of 学. The 'ch' in 'chung' (成): curl the tip of your tongue UP and BACK toward the middle of your roof, like a cat's 'ch'. The 'j' in 'jee' (只): same as 'ch' but unvoiced-start, tongue tip curled up-back." } ], "proverb": { "text": "邯郸学步", "literal": "Learning to walk in Handan", "means": "A man from the state of Zhao went to Handan to learn their elegant walking style. After trying to copy every step, he forgot how to walk at all and had to crawl home. Used to mean: imitating someone else so obsessively that you lose your own ability entirely.", "why_this_one": "The 'Handan walk' is the dominant AI paradigm (currently US-dominant: open models, research culture, compute advantage). The 'man from Zhao' is the side trying to replicate it. The 'uneven road' complaint is the sanctions / export-control / 'unfair playing field' rhetoric. I picked it over 田忌赛马 (horse-racing strategy) and 龟兔赛跑 (tortoise and hare) because the physical image of forgetting your own legs and crawling is funnier to say on camera, and it sits around the 60th–80th most common idiom, not in the top 30 a Chinese person reaches for first." }, "tones": [ { "tone": 1, "arrow": "→", "what_to_do": "High and flat. Start at the top of your pitch range and hold it there, like a sustained 'eee' on a high note. Don't let it drift down. Think: a surprised 'Oh!' held steady." }, { "tone": 2, "arrow": "↗", "what_to_do": "Rising. Start at mid pitch and slide UP, like asking 'Huh?' or 'What?' in English. End higher than you started. Don't start too low or you'll run out of room." }, { "tone": 3, "arrow": "↘↗", "what_to_do": "Dipping. Start mid-low, dip to your absolute lowest, then rise back to mid. A small 'U' shape. It takes about the same time as a 4th tone. Do NOT flatten it to a low tone. Think: a skeptical 'Huh…' that goes down then up." }, { "tone": 4, "arrow": "↘", "what_to_do": "Falling sharply. Start high and slam DOWN, like a firm 'No!' or 'Stop!'. Fast, decisive, no lingering at the bottom. Think: a door closing." }, { "tone": 0, "arrow": "·", "what_to_do": "Neutral. No pitch movement. Say it short, light, and unstressed, like the 'a' in 'a' or the 'to' in 'to'. Let it ride on whatever pitch the previous syllable left you at." } ], "hard_sounds": [ { "sound": "x [ɕ] (in 学 shweh, 嫌 shyen)", "what_to_do": "Say 'sh' (as in 'she') but pull the tip of your tongue DOWN behind your bottom front teeth so it does NOT touch the bumpy ridge behind your top teeth. The tongue is flat and low. If your tongue touches the ridge, you'll say plain 'sh' and the machine will mishear." }, { "sound": "ü [y] (the 'weh' part of 学 shweh)", "what_to_do": "Say 'ee' (as in 'see') but purse your lips forward as if you are about to whistle. Your tongue STAYS in the 'ee' position; only your lips change. In 'shweh', the 'weh' part is this sound: 'sh' (tongue tip down) + 'ee-with-rounded-lips', all one syllable." }, { "sound": "ch [ʈʂʰ] (in 成 chung)", "what_to_do": "Say 'ch' (as in 'church') but curl the tip of your tongue UP and BACK toward the middle of your roof. The tip should NOT touch the front ridge. It's a 'ch' made further back in the mouth, like a cat's 'ch' sound." }, { "sound": "zh [ʈʐ] (in 只 jee)", "what_to_do": "Same tongue position as 'ch' (tip curled up-back) but no burst of air. Say 'j' (as in 'jump') but with the tongue tip curled up toward the middle of your palate instead of flat against the front ridge." }, { "sound": "3rd tone dipping (in 比 bee, 只 jee)", "what_to_do": "This is a TONE, not a consonant, but it's the one you'll most likely flatten. Start at mid-low, go to your lowest, then rise to mid. A clear 'U'. If you just go flat-low, the machine may still catch it, but if you make it too short (under 200 ms) it will be missed. Give it time." } ], "notes": "For Quip 1 I traded the drier, more statistical '拒答率比正确率还高' (refusal-rate higher than accuracy-rate, which contains 率 lǜ and 确 què with the ü sound and 正 zhèng with retroflex zh) for the colloquial '拒答的比答对的还多' (the ones refused outnumber the ones got right). Same joke, zero ü, zero zh, zero ch. You should land this one cleanly. For Quip 2 I kept 学 (xué) because the 邯郸学步 reference is the whole joke and you can't swap it out. I expect Quip 2 is the one you lose: 学 appears twice and the x+ü combination is the single hardest sound in Mandarin for an English mouth. If 'shweh' comes out as 'shee' or 'shoo' the voice input will hear 数 or 树 and the sentence falls apart. Practise the tongue-tip-down 'sh' + rounded-lip 'ee' combo five times before you hit record. The 3rd tone on 只 (jee) in Quip 2 is a secondary risk: if you flatten it the machine may still parse the sentence, but the rhythm will sound wrong to a Chinese ear." }