Cantonese Text-to-Speech (TTS) System

                                                                                                        also on  Androidand  iPhone

New Super Natural Male Voice

Web Demo

Main Features


Screen Shots



Demo Download

System Requirements

Contact Us



The Cantonese Text-to-Speech (TTS) system introduced by PiTL provides a CD quality output from Chinese (Big5) Text to MP3.  The system consists of the technologies in Speech Signal Processing, Artificial Intelligence and Cantonese Linguistics that can transform a Text into native Cantonese Sounds. It is a Hong Kong style Cantonese speaking machine and can pronounce most of the local dialect (including native slang) of everyday language spoken by Hong Kong citizen.  The system is tested to read different style of literatures: Classics, Poems, Fictions, News, Jokes, Advertisements, and Movie Scripts and offer a high literacy of the read out.

This system also caters for the Visual Impair Persons who control the program with Key-strokes instead of Mouse clicks – all functions in the program have the corresponding short-cut key-strokes.

Mobile Devices enables people getting information on the go. Many people download Newspaper, Magazine, or e-Books to the Mobile Devices on demand. TTS technology enables a low cost and fast production of audio media and combining TTS and MP3, people can listen to the information at any time any where. Most important, no lighting, no glasses, no paper are required.

We also post a concept of e-Newspaper. Although Mobile Device claims to be very handy for mobile information, however, due to the limitation in the size of the screen, reading large amount of text from small screen is an eye-sight burning exercise. If we can put only Headings/Graphics on the screen and the details are left for the user to click the audio links and then listen, we can include lots of information in a pocket size mobile device screen.

“Point to the Titles, Listen to the details.”

PiTL believes that the merging of e-book, MP3 and TTS technologies will bring us a new information era.

Demo Version Download
Try out our demo at: speakd03.zip (9MB)

(Please read Installation.txt before install)


Purchase Full-version: HKD200 per license (single user)


System requirements:
You: Understand Cantonese, Can read Chinese
Your PC:

Microsoft Windows XP or later

Set to support Traditional Chinese version (Big5 Character set)

Intel P4 1GHz or above

256 RAM

Sound Card & Speaker

100M Free Hard-Disk Space in C: Drive


Main Features

True Voice                Sound Database is built on a high quality samples. This creates a more natural reading output. The MP3 files are written out at 64K encoding and is comparable to CD quality. In the future, more Sound Profiles will be created and will include different kind of expression and mood (such as: Happiness, Angriness, and Sadness). This is very difficult to be achieved by digital synthesis speech.


Variable                    By employing advanced DSP technologies, the reading speed can be adjusted at real-time without affecting the speech

Speed                      quality and pitch. Users can custom the reading speed to fit their preferences.


Rhythm                    Each sentence will be analyzed by the built-in AI mechanism and generated a reading tempo.

This will produce a more colorful sounding (especially fits for reading Chinese Poems).


Fast                         Real time reading. Batch processing of text files. (Individual file size can be up to 20K Chinese Characters). In 15 minutes, 20K Chinese Characters could be read to a total play time of 30 minutes MP3 files.( Intel P4 1GHz standard).


Accurate                  The system references to various Chinese dialect, linguistic and grammar dictionaries to handle the Polyphonic Characters  in Cantonese.  More than 90% accuracy can be achieved in reading all types of literatures ( News, Finance, Entertainments, Classics, Poems, and Fictions).


GB&BIG5                 Build in GB to BIG5 Character set converter allows the system to process all Chinese characters.


English                    Most of the Chinese TTS will only pronounce the English alphabets. Our system handles a mixture of both Chinese and English.

This is of importance for nowadays text information.

(English TTS is provided by the Microsoft internal TTS engine which is installed in Window XP).


Custom                    Flexibility is built in the system that allows sound editing. Users can select the correct sounding for a character before MP3 generation.

Sound                    This will increase the literacy and make it simpler to handle the complex Chinese Language.


Hong Kong              It is a Hong Kong style Cantonese speaking machine and can pronounce most of the local dialect (including native slang)

Style                  of everyday language spoken by Hong Kong people.  The system will support HKSCS (Big5 extension for Hong Kong character set). Comics and Conversation scripts will be read out in a natural way.


Screen Shots










Custom Sound

Usually applies in Surname

(=zang1=) ((=zang1=)先生的曾(=zang1=))


有人參(=sam1=)加入 (人參(=sam1=)-保品)



Sample Readings



12 9 星期二 16:20 更新







Fashion One 開倉﹗

尚有兩星期聖誕就殺到埋身。隨住經濟好轉市面的確興旺起來人人忙著蒐羅聖誕禮物也忙著裝身參加派對去。想掃貨又不想太破費﹖當然要抓緊開倉這些良機。集多個國際級品牌和大師設計的鞋店-Fashion One,於今日開倉。所有鞋履、手袋、New Style 圍巾、皮草及配飾等,全部低至三折起。當中還有今季 New Times 和 Party Mood 等靚鞋選擇。想去派對時有對靚鞋著(=zoek3=),記住時間去掃貨。



Hong Kong Style dialect and slang

發達啦(=laa3=)! 我中(=zung3=)左今期嘰(=ge3=)六合彩喎(=wo3=), 呢次打跛腳下半世都唔使憂囉(=lo3=)!


X(=ce1=),發白日夢咋, 唸(=lam2=)下(=haa5=)都冇咩問題架 (=gaa3=)嘛










Romance of the Three Kingdoms

選自: 三國演義

第一回: 宴桃園豪杰三結義,斬黃巾英雄首立功.





           一壺濁酒喜相逢。古今多少事,都付笑談中。 ——調寄《臨江仙》












Listen:sgyy01.mp3(Fast); sgyy02.mp3(Slow)



A trail on Intonation

In this new version (V2), intonation elements are added in. This will it make more colorful when reading poems.



                                                                                                                                                        (More) 更多... (Chinese Only)


Besides personal use, there are several areas of application:


E-Book Publishing                  

Publishers, Writers and Media can think about offering Audio Books or use it for Proof reading.



MP3 Audio lecture notes, Audio text book. Students can repeat a session for enforcing the memory.


Visual Impaired Person

The system has been sold to some VIPs. Their comments are “Those days of unnatural digital voice are gone. The system brings them more human




News, Weather, Finance updates, Speaking Web pages. Mobile Information Providers can make use of the system to create audio sources.


Mobile Devices

Mobile devices play music for Music lovers. How about “Information lovers” , “Gossip lovers”, “Book lovers” ? 

There are big markets out there to explore.


Learn Cantonese

Cantonese is one of the oldest dialects in China. Many characters in classics and poems have the corresponding Cantonese

Sounds. It is also the most popular dialect in Hong Kong.

Contact Us

Pacific i-Technology Limited (PiTL)

E-mail:      more@pitl.com

Tel:           (852)-2992 7108

Pacific i-Technology Limited (PiTL)    Email: more@pitl.com  TEL:(852)-2992 7108