The six tones
Every Cantonese syllable carries one of six tones, described by Chao tone numbers that trace the pitch from start to finish on a 1 to 5 scale. Three of them are flat, but at three different heights, and two of them rise, but from different starting points. That is why Cantonese tones are harder than Mandarin's: level matters, not just shape.
- Tone 1, high and flat (55): sing one steady high note at the top of your range. Example: si1, 詩, poem.
- Tone 2, high rising (25): start in the middle and climb to the top, like asking a question. Example: si2, 史, history.
- Tone 3, mid flat (33): sing one steady note in the middle of your range. Example: si3, 試, to try.
- Tone 4, low falling (21): start low-mid and drop to the bottom, like a sigh dropping away. Example: si4, 時, time.
- Tone 5, low rising (23): start near the bottom and climb to the middle. Do not climb to the top: that would be Tone 2. Example: si5, 市, city.
- Tone 6, low flat (22): sing one steady low note, below Tone 3 but not at the very bottom. Example: si6, 事, matter.
Cantonese Tone Trainer judges your pitch against the target's level AND shape. A low voice and a high voice are both judged against their own calibrated range, so the check works for any voice.
Calibrate your voice range first
Before practising, press Set my voice range and slide slowly from your lowest to your highest comfortable note, saying a steady "ahhh" while you slide. The app measures your range from that slide and grades every later utterance against it: 1.0 means the top of your range, 0.0 the bottom. This is what makes the three flat tones gradeable: Tone 1 sits near 0.9, Tone 3 near 0.6, Tone 6 near 0.3. Without calibration, a flat tone at any height could be any of the three. You can recalibrate any time in Settings, and your range is stored only on your device.
The 18 syllables
Three everyday syllables, each in all six tones, so you hear how the tone alone changes the word:
- si: si1 詩 poem, si2 史 history, si3 試 to try, si4 時 time, si5 市 city, si6 事 matter.
- fan: fan1 分 to divide; a part, fan2 粉 powder, fan3 訓 to instruct; to train, fan4 焚 to burn, fan5 奮 to exert; to strive, fan6 份 portion; share.
- jyu: jyu1 朱 vermilion; a surname, jyu2 主 master; main, jyu3 注 to pour; to note, jyu4 如 like; as, jyu5 乳 milk; breast, jyu6 預 in advance.
Tone pairs: 12 real words on the hard contrasts
Knowing single tones is only half the skill. In real speech, tones come in pairs, and the practice words are chosen to drill the confusable contrasts: the three flat levels, the two rises, and the low fall.
- Flat levels: do1 ze6 多謝 thank you (1-6), baai3 baai3 拜拜 bye-bye (3-3), sik6 faan6 食飯 to eat a meal (6-6).
- The two rises: nei5 hou2 你好 hello (5-2), hou2 noi6 好耐 a long time (2-6), dim2 gaai2 點解 why (2-2).
- With the low fall: zou2 san4 早晨 good morning (2-4), m4 goi1 唔該 please; excuse me (4-1), m4 hou2 唔好 do not (4-2).
- And three more: heung1 gong2 香港 Hong Kong (1-2), faan1 gung1 返工 to go to work (1-1), gei2 do1 幾多 how much (2-1).
Tone pairs unlock at 50 points, after you have the single tones down. The app splits your utterance at the gap between syllables and checks each half against its target tone. If the first syllable is right and the second is wrong, the hint says so: Second syllable, then the named fix.
How the live check works
Your browser samples the microphone and measures your pitch with autocorrelation, a standard signal-processing method that finds how often your voice waveform repeats. No model is downloaded and no audio is sent anywhere. Your pitch contour is measured against your calibrated range and compared to the six standard target shapes. A match needs a close fit and a clear margin over the next-best tone; a contour sitting halfway between Tone 3 and Tone 6 is honestly reported as unclear rather than guessed. Press Hear the target to hear any tone shape as a gliding tone placed inside your own range before you say it.
Entering tones: 7, 8 and 9
Cantonese is often described as having nine tones. Tones 7, 8 and 9 are the "entering" tones, heard only on syllables ending in -p, -t or -k. Their pitch contours match tones 1, 3 and 6 respectively; they are shorter because the syllable ends in a stop consonant. This app teaches the six core contours, which cover the entering tones too. A teacher or native speaker can show you the stopped endings.
Quiz, points, levels and badges
Quiz mode asks for 8 items in a random order: +10 points per match, +20 bonus for a perfect round. Levels run New Voice, Tone Learner, Tone Star, and Master of Cantonese Tones. Badges cover the first matched tone, practising all 18 syllables, practising all 12 tone pairs, a perfect quiz round, and practising on seven days in a row. Confetti and chimes celebrate wins, and both can be turned off in Settings. Practice stays the core: the games are there to pull you back for one more tone.
Your voice never leaves your device
Voice frames are analysed on your device and discarded. The page security policy blocks connections to every outside server. There is no account, no analytics, no recording, and no cookie. The built-in Offline proof panel shows a live log of every network request the page makes, lets you simulate a fully offline session, shows exactly what is stored on your device, and deletes it all with one button. The app works offline after your first visit. Pitch tracking approximates tones, and regional variation exists: the target shapes describe standard (Guangzhou/Hong Kong) citation tones. Native-speaker review is being sought in parallel, not as a gate.
Learning Mandarin too? Try the Mandarin Tone Trainer, the same live pitch feedback for Mandarin's four tones. Learning Vietnamese too? Try the Vietnamese Tone Trainer, the same live pitch feedback for Vietnamese's six tones. Learning Thai too? Try the Thai Tone Trainer, which coaches Thai's five tones the same private, on-device way. For sign language, Fingerspell teaches the ASL alphabet the same private, on-device way. Learning Japanese pitch accent too? Try the Japanese Pitch Accent Trainer, which coaches Tokyo pitch patterns with the same private, on-device feedback. Learning Taiwanese Hokkien too? Try the Taiwanese Hokkien Tone Trainer, which coaches all seven Hokkien tones the same private, on-device way. Learning Yoruba too? Try the Yoruba Tone Trainer, which coaches Yoruba's three tones the same private, on-device way.
This is a home practice tool, not a teacher, therapist or safety device. Everything runs on your device: no video, audio or images are sent anywhere. Feedback is automatic and can be wrong. For real instruction, learn from a qualified Cantonese teacher or a native-speaker community, linked on the Resources page.