People with serious speech impairments, such as physicist Stephen Hawking, may soon be able to shed their reliance on speech devices and sound more human-like by using unique, synthetic voices.
Every person, including the speech-impaired, has a unique 'vocal source', or buzzing sounds generated by the larynx. In normal speech, people are able to push the vocal source through the rest of the vocal tract.
U.S scientists have now built a program that creates customized voices for the speech-impaired. Patients are first asked to produce whatever sounds they can. Next, the program recruits voice donors of a similar age and build to the patient. The donors then repeat a series of short sentences covering all the sound combinations of their language in a three-hour recording session.
By using special software, experts are then able to create voices which combine the donor's natural speech and the patient's 'vocal source'. The synthetic voice allows patients to speak any word or sentence - even those not recorded by the donors.
VocaliD, the non-governmental organization set up by Rupal Patel of Northeastern University and Tim Bunnell of Nemours Alfred I duPont Hospital for Children, estimate that about 2.5 million Americans have severe speech impairments. Many patients with conditions such as Parkinson's Disease and cerebral palsy are unable to speak due to damage to the substantia nigra and cerebrum.