TECHNOLOGY
KEN DAVIES/CORBIS
A voice to call your own
into new words. “You probably wouldn’t recognise it as having come from the donor any more,” says Timothy Bunnell of the University of Delaware in Wilmington, who created The right synthetic voice can empower people who cannot speak ModelTalker and is also VocaliD’s co-director. Using this method, the group has built a handful of personalised voices. The impact it has on recipients is huge. As one anonymous user put it: “I was almost in tears when I first heard it and I can’t express what it means to know that, whatever happens to me, I will be able to communicate with my own voice.” But the process is slow, since surrogates must come to a studio to record for several hours. It takes at least 800 sentences to create a usable voice, and around 3000 for one that sounds relatively natural. VocaliD wants people all over the world to donate their voices so that what Patel calls their “voice bank” will have a whole range of speaking styles on tap. “If we were successful at being –Putting words in another’s mouth– able to do this data collection via the iPhone, we’d really get to synthetic voices built for them. capture the variation of voices utterances provide clues to what Aviva Rutkin When movie critic Roger Ebert that person’s speech might sound in the world,” she says. I AM locked in a quiet, carpeted lost his ability to speak due to The team hopes to encourage like – whether it’s high-pitched, room, listening to a robotic voice cancer, Scottish text-to-speech children to contribute their voices raspy or breathy. on my iPhone. When the voice company CereProc was able to by building a game around the A surrogate who is similar in pauses, I repeat after it: “Jo’s build a substitute that sounded recording process, which can feel age and the same sex is selected gentlemanly demeanour amused close to his own. to donate their voice. That person a little tedious at times. and set him at his ease.” But most people don’t have a “This is a significant step reads through several thousand This rather odd sentence, along rich supply of audio recordings at sample sentences, sourced from forward in using technology in a with a few hundred others, form their disposal to help stitch a new way that’s quite novel,” says David part of a test for a new phone app voice together. Generally, they are “It’s crucial that a synthetic Pisoni, director of the Speech voice fits like any other that will allow people to donate stuck with generic computerised Research Laboratory at Indiana prosthetic – it’s the only their voices to help other people. voices – think Stephen Hawking, University. It is important because means of interaction” VocaliD wants to use speech who uses an early synthesiser how someone speaks gives a recordings to create personalised called DECtalk. listener much more information synthetic voices for those who “For these individuals, this is about them than just the content classic books like White Fang, are unable to speak on their own. the only way that they interact of what they say, he says. The Wonderful Wizard of Oz and One day, my voice may be cut with people around them,” says “Those attributes of your The Velveteen Rabbit. up and customised for a person Rupal Patel, a speech scientist at speech tell the listener whether Then, using a software tool who needs a voice that sounds a Northeastern University in you are familiar or unfamiliar, called ModelTalker, the bit like mine. Boston and VocaliD’s co-director. male or female, from New surrogate’s voice is blended with There are millions of people It’s crucial that it fits, just like England, New York City or the the patient’s and stripped down with severe speech impediments any other prosthetic, she says. South. It tells you about the into the tiny units that make up because of a stroke, Parkinson’s So her group listens to the speech. Even a single vowel sound emotional and mental space of a or cerebral palsy, for example. limited sounds that her patients might be broken into two or three talker,” Pisoni says. “This is what In the past, a lucky few have had are able to produce. These makes you you and me me.” ■ parts that can then be assembled 22 | NewScientist | 22 March 2014