Published on Jul 11, 2005

Soon sufferers from dysarthria and others who have difficulties communicating clearly and completely will be able to communicate understandably and also control appliances using special tools and a newly developed speech-recognition system and speech-synthesis module.

Dysarthria is an impaired articulatory ability resulting from defects in the peripheral motor nerves or in the muscles that control speech. It is one of a family of motor speech disorders which includes neurological trauma and conditions such as cerebral palsy, a loss or deficiency of motor control with involuntary spasms caused by permanent brain damage present at birth. This condition severely restricts the ability to communicate with others.

To enable a person with dysarthria to communicate with others, speech technologies are essential to improve their communication and at the same time enable them to control assistive devices to operate home appliances and other things in daily life.

Realising that standard speech-recognition technology is not enough to enable communication by people with dysarthria, the National Electronics and Computer Technology Centre (Nectec)'s Assistive Technology Centre (Astec) has been working on a project to develop speech-enhancers as well as tools for people with these difficulties.

As a division of Nectec, Astec is dedicated to the research, development, design, and engineering of assistive technology and devices to help people with disabilities to access information and use computers effectively. Other projects are the write-and-talk word-processor for children; the talker, a speech output communication program for non-speaking persons; and Opal, a hand-held communication device for non-speaking persons.

Prakasith Kayasith, Astec's research assistant, said the project's main objective was to use speech technology to improve social communication and to control common electronic devices in houses, for example to open windows or doors, to turn a television on or off and, more importantly, to call for help.

The project consists of three subprojects.

Collaborating with researchers both from universities and hospitals, the team has been working on the development of a speech-training system, a speech-enhancer and a robust speech-recognition system for those with dysarthria.

The speech-training system will be a real-time visual feedback software package to assist speakers to improve the consistency of their vocalisation, while the speech-enhancer will be an electronic device designed to enhance the quality of sound. The device will be created to clarify a user's voice for easier understanding by listeners and by automatic speech-recognition systems in devices used by the speakers.

In the case of severe dysarthria, the development of the speech-recognition system will produce an attachment for other applications: for example it could be added to a speech-synthesis module or to electronic-device controllers to operate appliances, probably in the home.

Prakasith said the three-year project had four phases. Having started the project about six months ago, the team is in the first phase, building a speech database gathered from group of dysarthria sufferers. The work includes study of core speech patterns, speech-error analysis and a classification framework.

Astec has had cooperation from experts in different fields at Siriraj Hospital and Chulalongkorn University. Dividing the work into three steps, the team has joined with speech therapists and linguistic experts to find errors from examples of speech they have gathered. Physicists then study the acoustic elements of sample sounds to detect speech errors and process the signals to make clarifications.

"We expect to divide the groups with the disability into mild, moderate and severe. This will be a foundation for work on ways to improve their symptoms," he said.

Prakasith said that once the database had been completed, the information would be used in developing an automatic speech-recognition engine, which would be used to control electronic devices and home appliances.

Connecting the engine to a transmitter built in a computer, words spoken by a dysarthria sufferer are transformed into sound waves to be sent to independent hardware called environmental control units used to command and control devices.

"The hardware unit will work to interface commands sent from a computer to electronic devices and appliances using connecting wires or infrared signals. To control appliances such as televisions or radios we will develop a universal remote control to embed in the system," he said.

The development of the automatic speech-recognition engine will be used as a base for the development of the speech-enhancer and speech training. Nectec has funded this part of the development with about Bt4 million.

Although designed mainly to assist dysarthria sufferers, Prakasith said, the project developments, especially the speech-enhancer, could be used with groups of patients who have poor vocal-cord operation, such as senior citizens, since the device is intended to enhance the quality of speech.

Prakasith has also submitted the project for his doctoral degree at Thammasat University's Sirindhorn International Institute of Technology and received a scholarship from the Thailand Research Fund to continue his study on the project.

Suchalee Pongprasert

The Nation

suchalee@nationgroup.com