The University of Sheffield
Department of Computer Science

21 February 2007

Barnsley accent welcomed for unique project

CAST team uses local voices in communication aid

Researchers at the University of Sheffield are using the distinct Barnsley accent of local poet Ian McMillan to help them develop a unique hand held communication device for local people with speech disorders.


Mr McMillan has been working with researchers at the University on the Voice Input Voice Output Communication Aid (VIVOCA) project, which aims to develop a portable, (eventually body worn) communication aid for people with disordered or unintelligible speech.


As part of the project, which is funded by the Department of Health and involves collaboration with Barnsley Hospital NHS Foundation Trust, researchers have built a computer-generated voice which resembles Mr McMillan, using recordings of his speech.


The research team has previously shown that it is possible for a computer to recognise severely disordered speech, even if it is unintelligible to human listeners. The VIVOCA device will work by recognising such speech and then outputting clear speech which conveys the intended message. The researchers believe the output speech will be more acceptable to the user if it sounds natural and has a local flavour – the voice of Mr McMillan.


For instance if someone with a speech disorder was buying milk, they could speak into the device and ask for milk in their own voice, even if it was unintelligible. The device might then say “A pint of milk please” in the style of Mr McMillan.


This aid, which will consist of a hand-held computer (a PDA) and a wireless Bluetooth headset, similar to those used for hands free kits for mobile phone, is to help people who have difficulty in controlling the muscles involved in producing speech. This can happen after strokes or as a result of disorders such as Parkinson’s disease, Cerebral Palsy and other motor neurone diseases.


Professor Phil Green, Head of the Department of Computer Science at the University of Sheffield, who has been working on the VIVOCA project, said: “A person’s voice characteristics convey features of that person’s identity, such as what sex and age they are and where they are from. If the person is using the device as their primary mode of communication, it is important that the output voice is suitable to represent that person.”


He added: “We contacted Ian McMillan because his voice could provide that geographic identity for a local VIVOCA user. His professional experience as a voice artist also means he was able to make accurate, consistent and natural sounding recordings from read speech. Ian is the first of our ‘VIVOCA voices’. Eventually we would like to be able to present a client with a choice of male and female voices, and perhaps also adapt their chosen voice to resemble that of the client before their speech deteriorated.”


Ian McMillan is delighted to have been able to offer his voice to help people in the local community. He said: “'I'm happy, excited and flattered that my voice is going to be used to help others find their voice; for years I've wanted people to find their creative voice, but here's a concrete example of them finding their real voice!”


Professor Mark Hawley from Barnsley Hospital NHS Foundation Trust, who leads the VIVOCA project, commented: “We are extremely grateful to Ian McMillan for all his help with this device and for helping us tailor the VIVOCA aid to the specific needs of local users.”




Notes for editors

The VIVOCA project is funded by the Department of Health New and Emerging Applications of Technology (NEAT)Programme. The partners include: Barnsley Hospital NHS Foundation Trust, The Department of Computer Science at the University of Sheffield, The Department of Human Communication Sciences at the University of Sheffield and the Institute of General Practice and Primary Care in The School of Health and Related Research at the University of Sheffield.


The researchers involved in the project are all members of Clinical Applications of Speech Technology (CAST). CAST is a multidisciplinary research group made up of representatives from the University of Sheffield and Barnsley Hospital interested in applying speech technology in clinical areas such as assistive technology, electronic control systems and speech and language therapy.