RESEARCHERS at Sheffield University are working on a pioneering project to ‘bank’ a person’s voice to help them communicate if they lose the power of speech.
The experts believe they have found a way to give patients with degenerative conditions a synthesised version of their own voice - giving new hope to sufferers of illnesses like motor neurone disease and Parkinson’s disease.
Dr Sarah Creer, from the university’s School of Health and Related Research, is a research assistant on the project. The team has ‘proof in concept’ that the idea can work.
The work involves ‘banking’ a person’s speech and using the recordings to build a synthetic voice which sounds more like the patient’s natural voice. It can also be used for patients who know they are set to lose their voice, ahead of a planned operation such as a laryngectomy.
The team is working with universities in Edinburgh and Japan to develop the software which will allow them to turn it into a communication aid used by patients.
Dr Creer said: “We want to develop a communication aid for people whose voices have started to go, such as those with MND or Parkinson’s disease, the kind of conditions which cause progressive deterioration.
“We want to create a device which can output an intelligible version of their own voice, making it sound more like the person who is using it.
“This is important because by the time these patients start thinking about banking their voice, it could already be too late because the condition has started to take hold.
“We want to give people the opportunity to restore the bits that have deteriorated or gone wrong if they want to. We still need to do the research on how to put it all together from a technical point of view, but our aim is to create something that will give people the choice of holding on to their own voice.”
Some systems already exist and currently work as communication aids for such patients before they have started to deteriorate.
But the most popular device requires a patient to record 1,800 sentences and phrases from which any sentence a patient wants to speak can be generated. The local project would require a maximum of 200 utterances to be recorded.
“This is a huge difference in the amount of data required,” said Dr Creer.
“This is particularly important for people whose voices are beginning to deteriorate, who may be suffering other effects of the illness and for whom time is very valuable.
“There is currently a lot of rejection of communication aids because people do not associate with them at all - they feel it’s not part of them.
“We hope to make it easier for these patients to hold onto their own voice, which is such a label of identity and which can show so much about a person.”