The weekly column

Article 43, December 2000

Simulated Native English Speakers

By Guo Shesen

I am an English teacher in the English Department, Luoyang University, Henan Province, P.R China. We cannot invite more native English speakers to teach listening and pronunciation due to lack of finance. Our Chinese English teachers’ pronunciation is of course not as standard as the native English speakers’. It is a big problem to record the listening questions for the final examinations or a random quiz as we have more than 70 classes and only one American.

By chance I found the Microsoft Text to Speech Engine when I surfed on the Internet and, very importantly, it is absolutely free! This is really great news to me because I had heard before about some other paid TTS Engines which charged a lot and their English voices sound like a robot’s, not like a real man’s. Yes, Microsoft TTS will offer us free as many native English speakers as we need and will spell words or read the whole passage automatically with standard pronunciation at any time. It will solve the problem completely. [Editor's note: The Text-to-Speech engine is part of the Microsoft Agent, which can be downloaded from http://msdn.microsoft.com/workshop/c-frame.htm?/workshop/imedia/agent/default.asp ).

I downloaded the Microsoft TTS Engine to my computer and installed it. After examining important relevant class library files I was very happy to have found so many useful methods. As we know, the more methods provided in computer controls, the more flexibility and functions they will own. It was very possible for me to generate computer applications - simulated native English speakers.

I imported the files to Delphi 5.0 which had already been in my computer for programs to teach and immediately added the relevant unit to the uses part. I placed several buttons on the form, whose functions were to trigger correspondingly Read, Read Previous Sentence, Read Next Sentence, Pause/Continue and Stop, similar to pressing buttons on a tape recorder, which serve separately as functions of Play, Rewind, Fastforward, Pause/Continue and Stop.

Designing like this is actually for those who are familiar with operations of recorders and unfamiliar with computers, complying with the principles of human design and human engineering.

When I entered English words or digits and clicked the Read button, the standard and fluent English voiced with perfect intonation was read out. I was shocked when it read words I invented randomly as closely and as intelligently as a real man did. Yes, there are no words it does not speak, or rather there are no characters it does not read. Even a real and live native English speaker could not know as rich a vocabulary as it can. It is a great reader.

To control the reading speed I placed a trackbar on the form. This is really important to people who study English as a second language and to students who train for improvement of listening and speaking. I think this is revolutionary as the recordings we own in our department are all tapes or video tapes which cannot regulate the reading speed, that is, the reading speed of a tape is fixed, we have to press the button of audiorewind again and again to catch some difficult words which are read fast in the sentence so that they can be understood. Now we may use this magic method to set the reading speed as fast or as slow as we like to make the students understand better. It is a great speed controller.

To my surprise this TTS can jump reading and it can stop to read the next sentence correctly and immediately when it is reading. It will be difficult for a real man to do so and to a certain degree MS TTS is better than native live English speakers for it is never tired of speaking and repeating and jumping. It is a great jumper.

When I clicked the Read Previous Sentence, it would jump back to the previous sentence and immediately read. Again a great jumper.

It never complains of pausing and continuing. It will locate exactly where it paused and continue reading until the end of the passage as my finger flips on the Pause/Continue button. A real native speaker could envy its patience and energy.

When clicking the Stop button it will stop reading and wait for your next command, never going back to offices or houses for drinking or sleeping.

I succeeded in running it in Delphi and compiled it into an .exe file and copied it to other computers. They all became native English speakers who are never tired of speaking and repeating standard and fluent English at any time and place and need not be paid.

It is great.

But it is not everything. It has its own disadvantages. Sitting at the computer, I am still a bit dissatisfied with the super native English speaker with artificial intelligence. If she had human feelings……

Below is the picture of the main window of my Text-to-Voice program into which some editing functions have been integrated. I show it here only for exchange of new ideas with English teachers whose offices are in need of native English speakers or are probing into reforms of English teaching and not for any commercial purposes. Those who are interested in this application are welcome to visit my website at http://guoshesen.533.net for more details.

Picture of main window of my Text-to-Voice:

About the author

Guo Shesen (guoshesen@21cn.com ) is an English teacher in the English Department, Luoyang University, Henan Province, P.R China. For more information about Guo and his works, please visit his websites:

Homepage: http://guoshesen.533.net

Guestbook: http://guoshesen.abc.yesite.com

Also: http://enlforum.51.net

Questions or comments about this week's article? Why not post them on our Discussion Forum