Interaction Designer
CantoVoice - Cover Image.png

Canto Voice


Canto Voice



Cantonese is spoken by over 80 million people worldwide. However there is a growing divide between Cantonese speakers and their families. The goal is to understand these divisions, design a solution, and ship a real developed product with immediate benefits.



This project resulted in research findings that then informed the design and shipment of an app. Research found that there are multiple reasons Cantonese speakers may be distanced from relatives. Though many issues were systemic, there’s also some that are easily solvable within our short timeframe. One core issue discovered was the communication barrier between two main archetypes; 1. Senior family members who have a good understanding of the language but are often not tech literate, and 2. More tech-literate folks who can only speak the language, but can’t read or write (typically grew up in a country where Cantonese is not prominent) .

Cantonese has been largely forgotten in the digital era. It is unsupported by services like Google Translate, and Duolingo. Even when Cantonese is supported, it remains unstable and glitchy on popular platforms like iPhones.

As a result of the research and design process (see below), “Canto Voice” was created. It’s an app that aids day-to-day communication by helping less familiar family members understand, pronounce, and send messages in Cantonese with the aid of a unique romanization engine. (Romanization is the act of writing the pronunciation of a character.) “Canto Voice” has been designed and released as a beta. (It will also be added to an earlier communications project;, an instructional site which assists 1,800+ users every month.)


Try out the current (beta) version of “Canto Voice” on the iOS App Store.




Research & Design

To better understand how Cantonese is utilized, secondary research (quantitative and qualitative), observations, and user interviews were conducted. Competitive research also helped inform where there were opportunity spaces (of which there are many).

Early sketches served as research probes. They facilitated in the critique and prioritization of features whilst refining the target audiences of the app.

Some flows were sketched out in low-fidelity. However, upon doing so, it was realized that such a utilitarian app would be more than adequately served at a higher fidelity.

Some Information Architecture (IA) was drafted in Numbers.

Design Documentation was done with High Fidelity sketches using iOS Design patterns. These were also used in user testing.

Some interesting insights gained from this included the: educational benefits of tokenizing Chinese characters, and that default iOS design patterns could benefit from additional affordances.


Prototyping & Development

Development was done in parallel with UI Design, meaning High-Fidelity Visual prototypes could be tested alongside working Interactive prototypes. Interactive prototypes were user tested in comparison with Microsoft, Apple, and Google offerings. People were pleasantly surprised by the accuracy and relative ease of interaction offered by the app.

Evolution of textual presentation of Cantonese Chinese Transliteration.

“Canto Voice” is a full iOS application, as well as an iMessage application. It has standard capabilities like speech-to-text, text-to-speech, and can translate between Cantonese in Traditional Chinese, Cantonese in faux Simplified Chinese, and English. These are done using a mix of standard Microsoft, Apple System and Apple NLP APIs. Explicit care was also taken to minimize and visually communicate load times on the iMessage app.

We also use our own evolving Cantonese romanization database (currently Jyutping), beginning with a mix of online resources and community input. This allows as to tokenize Chinese text and transliterate Cantonese on-device, and in real time.


Future Work

“Canto Voice” is scheduled for continuous improvement, through cycles of Research & Evaluation, Prioritization, Design, and Release. Whilst iMessage was the most popular platform for our users (considering both Android and iOS users), WhatsApp and WeChat also saw relatively high usage. Immediate next steps includes expanding and user testing on other platforms, additional features like interpreting audio and images offline, as well as removing the Translation limit by commercializing the app and introducing a paid tier.

In our research, there are still many systemic issues that create barriers between Cantonese speakers and their families. This project was geared towards designing something that was quick and had real (shippable) impact. In the long run, solutions could be investigated that improve exposure to Cantonese through the availability of educational resources and cultural artifacts. “Canto Voice” is simply one first step towards eroding communication barriers in order to bring families closer together.