The Development of Synthetic Speech Aids for Patients With Acquired Disability

Keating, David (1988) The Development of Synthetic Speech Aids for Patients With Acquired Disability. PhD thesis, University of Glasgow.

Full text available as:
[img]
Preview
PDF
Download (9MB) | Preview

Abstract

Patients suffering from a variety of speech disorders can benefit from synthetic speech. This study concentrates on the dysarthric patients with acquired speech loss as these patients have intact intellect and are more likely to benefit from synthetic speech. The physical skills of these patients vary enormously and their needs and situations are different. The main part of this work is concerned with the design, development and evaluation of a range of speech aids to meet these varying needs and skills. Three methods of speech synthesis are used and their performance has been investigated by using a Diagnostic Rhyme Test to measure the intelligibility of individual words. The results of this trial showed Adaptive Differential Pulse Code Modulation (ADPCM) to be more intelligible than Linear Predictive Coding (LPC), both these methods being more intelligible than constructive synthesis. A further trial was conducted to measure the speech quality of phrases produced by the synthesisers. This showed listeners preferred listening to phrases constructed of LPC words than to phrases generated using Phoneme based synthesisers. Phrases with mixed LPC and constructed words were preferred to phrases of constructed words. The devices that were developed use different methods of synthesis and the choice of method was guided by these trials. The Pocket Speech Aid is a rapid access limited vocabulary communication aid which uses ADPCM synthesis. Direct selection is the method used to give users access to eight phrases. The Pocket Speech Aid has been very successful in practice. When used as a telephone aid eight out of ten patients increased their communication ability and when used as a conversation prompter ten out of fourteen patients were able to steer the direction of real time conversations. This device has generated a great deal of interest from other centres and the demand for the device which is currently being manufactured confirms that it has a role to play in assisting those with communication difficulties. The Macleod Unit was named after a remarkable patient suffering from Motor Neurone Disease who realised his speech would soon be lost and had the foresight to select a vocabulary and record the words on a cassette recorder. His 625 word vocabulary was transferred to the speech aid which uses an encoding method of word selection. Clinical feedback showed the device to be of benefit for this highly motivated individual but was less successful for other patients in this group who found the cognitive effort to select codes too great. An unlimited vocabulary device based on the commercially available VOTRAX which uses constructive synthesis was developed but this device was rejected because of the robotic sounding voice. A further unlimited vocabulary device prototype, the Uvocom, was designed to improve the speech quality and to investigate if there is a need for an unlimited vocabulary. The Uvocom uses a core vocabulary of 1000 LPC words and uses Phoneme back-up for words not stored in the core vocabulary. Trials with the Uvocom have indicated that quality speech in an unlimited vocabulary device is likely to benefit a small number of patients who have the physical skills to operate such a device. Finally, some indication is given of the directions in which future work could progress based on the proven success of the Pocket Speech Aid.

Item Type: Thesis (PhD)
Qualification Level: Doctoral
Keywords: Biomedical engineering, Speech therapy
Date of Award: 1988
Depositing User: Enlighten Team
Unique ID: glathesis:1988-77766
Copyright: Copyright of this thesis is held by the author.
Date Deposited: 14 Jan 2020 11:53
Last Modified: 14 Jan 2020 11:53
URI: http://theses.gla.ac.uk/id/eprint/77766

Actions (login required)

View Item View Item

Downloads

Downloads per month over past year