03.10.16
STAR: speech technology for articulation rehabilitation
Source: NHE Sep/Oct 16
Dr Stuart Cunningham from the Department of Human Communication Sciences and Centre for Assistive Technology and Connected Healthcare at the University of Sheffield discusses the progress on an innovative project which is using speech and language technology for articulation rehabilitation.
Speech and language technology, and particularly automatic speech recognition, is being increasingly used in therapies and interventions for many different conditions. This reflects the growing everyday use of speech interfaces with smartphones and tablets.
One area where speech recognition technology could make a big difference is in helping people with speech disorders.
Traditionally, articulation therapy is delivered by speech and language therapists, and is used to treat clients who have difficulties producing speech sounds correctly. It involves the client repeatedly speaking the sounds or types of sounds that they have difficulty with. Through repetition, and the expert feedback of their therapist, the client slowly improves the accuracy of their speech productions. This can lead to meaningful improvements in the intelligibility of their speech, and hence their ability to communicate independently with others. Unfortunately the repetitious nature of the practice is crucial to successful outcomes for the client, but this obviously makes it time-consuming and expensive to deliver.
Many speech difficulties can require some articulation therapy, but two groups of clients most frequently identified by speech and language therapists are people who have had a stroke and developed a condition known as dysarthria. The other group is children with hearing impairment. In both these groups the repetitious practice and feedback can help clients either relearn how to say difficult speech sounds, or learn to more consistently produce a specific speech sound.
STAR project
To address the difficulties in delivering articulation therapy, the STAR (Speech Technology for Articulation Rehabilitation) project, funded by the National Institute of Health Research, is developing an app based on novel speech recognition technology.
Researchers at Barnsley Hospital and the University of Sheffield have developed automatic speech recognition technology that can objectively ‘score’ speech productions. Research has shown that feedback based on these scores can be used to help people modify their speech productions.
Programmers at Therapy Box Ltd, an award-winning developer of communication apps, are currently developing the app.
The app will prompt the client to produce a word either visually, or by playing an example for them to imitate. Then, every time they speak the word, visual feedback is given on how close what they said is to the target specified by their therapist.
Motivational games
In the app, these speaking exercises will be presented as motivational games. Clients could then try and achieve a particular target in their practice sessions – or they may have targets set by their therapist in terms of number of repetitions or a particular score to achieve.
Speech and language therapists
Dr Stuart Cunningham from the Department of Human Communication Sciences and Centre for Assistive Technology and Connected Healthcare at the University of Sheffield discusses the progress on an innovative project which is using speech and language technology for articulation rehabilitation.
Speech and language technology, and particularly automatic speech recognition, is being increasingly used in therapies and interventions for many different conditions. This reflects the growing everyday use of speech interfaces with smartphones and tablets.
One area where speech recognition technology could make a big difference is in helping people with speech disorders.
Traditionally, articulation therapy is delivered by speech and language therapists, and is used to treat clients who have difficulties producing speech sounds correctly. It involves the client repeatedly speaking the sounds or types of sounds that they have difficulty with. Through repetition, and the expert feedback of their therapist, the client slowly improves the accuracy of their speech productions. This can lead to meaningful improvements in the intelligibility of their speech, and hence their ability to communicate independently with others. Unfortunately the repetitious nature of the practice is crucial to successful outcomes for the client, but this obviously makes it time-consuming and expensive to deliver.
Many speech difficulties can require some articulation therapy, but two groups of clients most frequently identified by speech and language therapists are people who have had a stroke and developed a condition known as dysarthria. The other group is children with hearing impairment. In both these groups the repetitious practice and feedback can help clients either relearn how to say difficult speech sounds, or learn to more consistently produce a specific speech sound.
STAR project
To address the difficulties in delivering articulation therapy, the STAR (Speech Technology for Articulation Rehabilitation) project, funded by the National Institute of Health Research, is developing an app based on novel speech recognition technology.
Researchers at Barnsley Hospital and the University of Sheffield have developed automatic speech recognition technology that can objectively ‘score’ speech productions. Research has shown that feedback based on these scores can be used to help people modify their speech productions.
Programmers at Therapy Box Ltd, an award-winning developer of communication apps, are currently developing the app.
The app will prompt the client to produce a word either visually, or by playing an example for them to imitate. Then, every time they speak the word, visual feedback is given on how close what they said is to the target specified by their therapist.
Motivational games
In the app, these speaking exercises will be presented as motivational games. Clients could then try and achieve a particular target in their practice sessions – or they may have targets set by their therapist in terms of number of repetitions or a particular score to achieve.
Speech and language therapists can use the information collected by the app when setting new therapy targets for the clients. The information could also be used to monitor the progress of clients between their therapy appointments. Generally, using a computer-supported therapy tends to increase the amount of practice that people do.
One of main innovations of the STAR app is the ability of the app to learn about an individual client. For each activity that a therapist creates with a client they are able to train the app to identify the target sound and the mistakes or incorrect sounds that client might make. In this way the app will be able to score the speech of the client based directly on the targets identified by the therapist.
We are currently still in the development phase of the project. Before the project ends, we will be evaluating the final prototype with clients and therapists in the NHS. If the evaluation is successful we hope the app will be available to clinicians and the public within the next 12 months.
The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health.
For more information
W: www.catch.org.uk/current-project/star
Tell us what you think – have your say below or email [email protected]