Loading

Best Cantonese Apps in 2026: A Polyglot's Honest Reckoning with a Broken Market

Cantonese has 85 million speakers. More than Italian, Korean, or Vietnamese. Yet when you search "learn Cantonese app" in 2026, the results are mostly lying to you. Most apps are repackaged Mandarin courses with the word "Cantonese" pasted on the marketing page. Some are pinyin-only flashcard apps that have never seen a Jyutping tone mark. The rest are abandoned 2014 hobby projects with fewer than ten lessons.

I have been learning Cantonese since my first trip to Hong Kong years ago. I speak 11 languages and Cantonese is one of my stronger ones, mostly because I worked harder on it than on any other language, because the resources are that thin. I built Mynago partly out of frustration with this exact problem. The honest count of apps that actually teach Cantonese, with native Hong Kong audio and a real curriculum, is six. Maybe seven if you are generous. The rest are noise.

This post is the reckoning. Why the market is broken, which apps are real, which are frauds, and what to do given that the venture capital that built Mandarin learning skipped Cantonese on purpose. The funder and political reality matters here. It is unusual to discuss VC dynamics in a language post. I am doing it because it is the single biggest reason the tooling looks the way it does, and learners deserve to know.

The honest disclosure first: I am the founder of one of the six apps I list as real. That conflict shapes how you should read everything below. I have made the case for Mynago on its merits and named the other five apps that compete with it without trying to make them sound worse than they are. The bio is on LinkedIn and as @langaholic if you want to verify the polyglot claim or the Hong Kong time.

The structural reason Cantonese has no Duolingo

Three reasons, in order of severity.

Six tones, no agreed romanization. Mandarin has Pinyin. One system, taught everywhere, every learner uses it. Cantonese has Jyutping, Yale, Sidney Lau, and Cantonese Pinyin all fighting. A serious app has to pick one (Jyutping is correct in 2026) and reject the others. Most apps cannot be bothered, so they ship inconsistent transcriptions or skip romanization entirely.

No proficiency exam. Mandarin has HSK. Cantonese has nothing equivalent. No standardized test means no curriculum scaffolding, no level-based marketing, no clear stopping points. App designers are allergic to that ambiguity.

Hong Kong is small and rich. That sounds backwards but it matters. Most Hong Kongers already speak English. The "I need to learn Cantonese for tourism" market is tiny because tourism in HK works in English. The "I need it for business" market is tiny because business works in English or Mandarin. The real audience is heritage speakers, partners of Hong Kongers, and a small group of polyglots and culture nerds. Not a venture-scale market.

There is also a political layer that nobody on app-store optimization blogs talks about. Cantonese in mainland Guangdong has been quietly squeezed out of schools and television over the last decade. Younger Cantonese speakers in Guangzhou code-switch heavily into Mandarin. Mainland VC firms have zero appetite for funding a language tool that smells like soft cultural resistance. Hong Kong VC mostly funds fintech. The natural homes for Cantonese app capital are exactly the rooms where the topic is uncomfortable.

The result: the apps that exist are passion projects, dictionaries with a Cantonese add-on, or generalist apps that bolted on a Cantonese track to fill out their language list. There is no Duolingo for Cantonese. There never will be. The realistic bet: Cantonese tooling continues to be built by small, motivated teams (often Hong Kongers in the diaspora) and by polyglot-founded apps where Cantonese is one of many supported languages and the founders care about getting it right. Mynago is in the second bucket. The other apps below are mostly in the first.

The six apps that actually exist

Ranked from most useful to least. Every one used personally, in Hong Kong cafes, on the MTR, or sitting at home practicing tones out loud while my neighbours wonder what I am doing.

1. Pleco (with Cantonese add-ons)

Not a course. The single best Chinese-language dictionary ever built, period, and the only reason a Cantonese learner can function in 2026. Free base app. Then you pay once for the Cantonese pronunciation pack (Jyutping on every entry) and the Cantonese audio pack. Buy both. They are inexpensive and you will use them for the next decade.

Pleco wins because Michael Love has run it for almost 20 years as a labor of love. It is what every other Chinese dictionary wishes it were. Limitation: it is a dictionary, not a teacher. The default views push Mandarin pinyin and you have to dig into settings to flip Jyutping to the front. Once you do, the dictionary handles Cantonese-specific entries and the OCR camera works on traditional script in HK signs. Spend the ten minutes to configure it the first time. The "what do I use instead of Pleco" question keeps coming up because of this discoverability problem, not because Pleco is wrong.

2. Mynago

I built this. The Cantonese course teaches in real Hong Kong scenarios (cha chaan teng ordering, MTR navigation, dim sum etiquette, code-switching with English) using native Cantonese audio and Jyutping-first scripts. Most Cantonese learners I hear from already study Mandarin or have Mandarin in their background, which shaped how the lessons handle the transfer. We explicitly call out where Mandarin instinct will mislead you on tones and final consonants.

The honest pitch: Mynago is the structured daily spine when CantoneseClass101 feels too podcast-shaped and Glossika feels too brutal. Limitation: same as anything else. Not a substitute for a tutor or living in Hong Kong. I built this app for a learner who looks like me, which is a polyglot trying to add Cantonese after Mandarin. If you are a heritage learner trying to recover the Cantonese your grandparents spoke, you will get value, but the angle is more Mandarin-bridging than heritage-recovery.

3. CantoneseClass101

The Innovative Language network's Cantonese product. Audio-first, podcast-format lessons with native speakers, dialogue breakdowns, and full transcripts. The library is genuinely large, beginner through advanced, and the lesson notes are good. The reason it is not number one: the upsell structure is exhausting and some of the back catalogue feels like 2012 audio quality. Use the free tier to test, pay for the basic plan if it clicks.

4. Glossika

Glossika is the Cantonese sentence-flood gym. You hear thousands of natural sentences, repeat them, and let your brain absorb the patterns. After two months of daily Glossika I noticed comprehension changes that vocab apps cannot produce. Not for absolute beginners. If you do not know basic vocabulary and tones, the firehose drowns you. Come back to it once you have three months of Mynago or CantoneseClass101 under your belt.

5. Drops

Drops actually has a dedicated Cantonese course, not just a Chinese pile-on. Vocabulary only, swipeable, beautiful illustrations, native audio with Jyutping. Five-minute sessions on the free tier are either restrictive or merciful depending on your relationship with phones. Drops is great for the first 800 words and useless after that. Treat it as a vocabulary pre-load before serious lessons.

6. Pimsleur Cantonese

The dinosaur of audio courses, and shockingly the Pimsleur Cantonese track is legitimate. Audio-only, call-and-response, no characters, no reading. What it does well is drill your ear for the six tones during a commute or a walk. What it does badly is everything related to reading, writing, and modern Hong Kong street slang. Use it only as ear training in the first three months.

Honorable mention. Anki with a community Cantonese frequency deck is essentially free spaced-repetition vocabulary for life, but it is a system, not an app, so I am not counting it. If you are willing to build your own workflow, Anki plus Pleco plus Mynago is a complete stack.

New entrant worth a look. HelloChinese added a Cantonese track in late April 2026. Early. Lesson count is thin compared to their Mandarin offering. Foundations are clean and the audio quality is up to their usual standard. Worth a look if you bounced off ChineseSkill or Glossika. The track has the Mandarin-app-translated-to-Canto smell on a few units, but most of it feels purpose-built. I will revisit in three months when the catalogue grows.

The four apps that pretend to teach Cantonese and do not

Naming these because the SEO ecosystem rewards them and learners waste months on them.

Duolingo. No Duolingo Cantonese course. They have Mandarin. That is it. Articles claiming "Duolingo for Cantonese" are SEO chum. Stop searching for it.

Rosetta Stone. No Cantonese product. Their Mandarin course exists but does not generalize, and they have no roadmap toward Cantonese that I have seen.

Babbel. Same story. No Cantonese. Babbel is a European-language company. They have never shipped a Chinese-variant product worth noting.

Generalist apps that claim "Mandarin and Cantonese" support. A common pattern: the Cantonese mode is the Mandarin curriculum with Jyutping captions auto-generated below the same Beijing-recorded audio. You hear Mandarin pronunciation while reading Jyutping. This trains your brain to associate the wrong sounds with the right romanization. Genuinely worse than no app. If you cannot hear a Hong Kong voice in the lesson, walk away.

What to actually do given the market

No single app finishes the job for Cantonese. The market is too thin. The path that works in 2026 is a stack, not a savior app.

Foundation (months 1 to 3). Mynago daily for structured lessons. Pimsleur on the commute for tone drilling. Pleco with both Cantonese add-ons as your dictionary reflex. Drops for the first 800 words if you like gamified vocabulary.

Building (months 4 to 12). Mynago continues as the spine. Add Glossika for sentence input. Add Anki with a community Cantonese deck for retention. Switch Pimsleur for CantoneseClass101 to get more variety in voices and topics. Start watching TVB dramas with Cantonese subtitles even if you understand 15%.

The non-app layer, the part that actually creates fluency. Get a tutor on italki who lives in Hong Kong. Forty-five minutes twice a week, in Cantonese, no English fallback after week three. This is where most of your real ability comes from. Apps prepare you for tutor sessions. Tutor sessions are the ability itself.

A word on tutors specifically: I am bullish on them for Cantonese in a way I am not for Spanish, where the resource ecosystem is so deep you can self-study to B2 without a human. In Cantonese, the lack of polished apps means a tutor is not a luxury. It is the missing piece of the curriculum. A good HK-based tutor on italki runs 15 to 25 USD per hour, cheaper than a single month of most language app subscriptions, and provides ten times the value for tone correction alone.

Immersion content for free. RTHK radio streams in Cantonese all day. TVB dramas on Viu and YouTube. Hong Kong YouTubers like Cantonese with Brittany and Cantolounge. The point is not to understand on day one. The point is to put your ear in the language for hours every week so the prosody, rhythm, and tonal contours soak in.

If you speak Mandarin already, you have a real reading shortcut. Characters transfer (adjust to traditional script). Formal-register vocabulary transfers. Tones do not transfer at all. 世界 is shìjiè in Mandarin and sai3 gaai3 in Cantonese. The mapping is unpredictable. Treat pronunciation as a complete reset. I write more about this tradeoff in Mandarin vs Cantonese: career vs soul.

A few things to push back on, since they come up in every learner forum. Cantonese is not dying. The pressure inside mainland China is real but Hong Kong, Macau, and the diaspora keep the language vibrant, and the cultural output (film, music, online video) is in better shape than it was a decade ago. Cantonese is not just spoken Mandarin with weird sounds. The grammar diverges in real ways, particularly the sentence-final particles (嘅, 啦, 囉, 咩, 㗎) that carry mood and tone and have no Mandarin equivalents. And traditional characters are not optional. Hong Kong, Macau, and most overseas Cantonese communities use traditional. If your character base is simplified-only from a Mandarin background, plan a few weeks to bridge.

If you are physically in Hong Kong: the 90-day plan

Different rules apply when you are physically in HK. The immersion budget jumps from "limited podcast time" to "every cab ride, every wet market trip, every breakfast." The plan changes accordingly.

Days 1 to 30. Mynago daily for 25 minutes. Pleco add-ons installed before you land. Pimsleur on every MTR ride. Order in Cantonese at cha chaan tengs even when staff switch to English. Point at the Chinese on the menu and read the Jyutping out loud from your phone. You will be bad at it for two weeks. Do it anyway. Get a tutor on italki who lives in HK by the end of week one.

Days 31 to 60. Glossika joins the rotation. Tutor sessions move to twice a week. Force yourself into one extended Cantonese-only conversation per day, even if it is small talk with a building security guard about the weather. Watch one TVB drama episode per week with Cantonese subtitles. Start reading the Apple Daily archive headlines (traditional script) with Pleco's reader function.

Days 61 to 90. CantoneseClass101 replaces Pimsleur. You are past the audio-only phase. Tutor sessions go to 60 minutes. Take a weekend trip to Guangzhou or Macau for a different Cantonese accent. Start a journal in romanized Cantonese (Jyutping is fine, characters are bonus). By day 90 you will handle taxi negotiations, restaurant orders, and small talk without panicking. You will not be fluent. You will be functional, which in a city where everyone speaks English is more than 95% of expats ever bother to become.

The thing nobody tells you about Cantonese in 2026: because the bar is so low, even modest competence is met with real warmth. People notice. They stop switching to English. They tell you their actual name instead of their English name. The language opens doors that Mandarin does not in the city where Mandarin is the bigger language. That is the whole reason to bother.

The reader question I get most

"I am moving to Hong Kong for work, do I learn Mandarin or Cantonese first?" The honest answer depends on whether your work language is English-plus or Mandarin-plus.

If your office runs in English with occasional Mandarin meetings, learn Cantonese. The professional network speaks Mandarin. The social network, the city itself, the doors that open to friendship all speak Cantonese.

If your office runs in Mandarin with English fallback, Mandarin first, Cantonese second, and accept that the second one will take longer. The reverse order locks you out of casual social Hong Kong in a way Mandarin learners do not realize until year two.

The market for Cantonese apps is small. The market for human connection in Cantonese is enormous. Choose your tools accordingly. It is not getting smaller either, which is a quiet positive after a decade of the opposite trend.

Keep reading