How Do You Construct A Voice?
STEPHEN HAWKING: Where did we come from? What is the future of the human race? Are we alone in the universe?
GUY RAZ, HOST:
So this is a voice you probably know - physicist Stephen Hawking, of course. But the weird thing about this voice is that it's not just his voice. This is the exact same computerized voice that's used by thousands of people around the world who cannot speak on their own. And we know this at Stephen Hawking because, well, he's Stephen Hawking.
RUPAL PATEL: But there's little girls and older women and lots of other people who are using that voice, even though he thinks of that as being his voice, right. And so do the other people who are using those voices. They think about them as being their voice.
RAZ: This is Rupal Patel.
PATEL: I'm a speech scientist and a professor at Northeastern University.
RAZ: And Rupal has figured out a way to create new voices, customized voices for people who use synthetic speech, people whose actual voices, for whatever reason - autism or cerebral palsy or stroke - can't speak. And the story of how it all came about - Rupal explains from the TED stage.
(SOUNDBITE OF TED TALK)
PATEL: This lack of individuation of the synthetic voice really hit home when I was at an assistive technology conference a few years ago. And I recall seeing a little girl and a grown man having a conversation using their devices.
UNIDENTIFIED SPEAKER #1: Nice to meet you
UNIDENTIFIED SPEAKER #2: Nice to meet you, too
PATEL: Different devices.
SPEAKER #1: How long have you been coming to this conference?
PATEL: But the same voice.
SPEAKER #2: This is my first year.
PATEL: I looked around, and I saw this happening all around me.
UNIDENTIFIED SPEAKER #3: I've never been there.
UNIDENTIFIED SPEAKER #4: Where is the registration table?
UNIDENTIFIED SPEAKER #5: What time should we meet back at the hotel?
PATEL: Voices that didn't fit their bodies or their personalities.
UNIDENTIFIED SPEAKER #6: Where are you staying?
UNIDENTIFIED SPEAKER #7: How long are you in town?
UNIDENTIFIED SPEAKER #8: Our flight was late coming in from Detroit.
PATEL: We wouldn't dream of fitting a little girl with a prosthetic limb of a grown man. So why then the same prosthetic voice? It really struck me, and I wanted to do something about this. I'm going to play you now a sample of two people, actually, who have severe speech disorder. They're saying the same utterance.
UNIDENTIFIED WOMAN: (Unintelligible).
UNIDENTIFIED MAN: (Unintelligible).
PATEL: You probably didn't understand what they said, but I hope that you heard their unique vocal identities. I wanted to find out how we could harness these residual vocal abilities and build a technology that could be customized for them, voices that could be customized for them. So we decided to do exactly that.
RAZ: With a girl named Samantha. Samantha has a very rare speech disorder, which makes it impossible for her to speak. But she can make sounds, vowel-like sounds, which turned out to be enough of a sample for Rupal and her team of researchers.
(SOUNDBITE OF RECORDING)
PATEL: That sound is just what Samantha can produce. That's her ah. And from that, we can gather the pitch of her voice.
(SOUNDBITE OF RECORDING)
PATEL: The quality of her voice, meaning things like is it raspy, is it breathy.
RAZ: So you are hearing all of these things. I mean, she can't create words or sentences, but she can make some basic sounds.
PATEL: Right. And we take what we can from those vocalizations and use them in the process for - we generate a voice for her that sounds like her.
(SOUNDBITE OF TED TALK)
PATEL: So how do you go about building this voice? Well, you have to find someone who's willing to be a surrogate. For Samantha, her surrogate came from somewhere in the Midwest - a stranger who gave her the gift of voice. Being a surrogate donor only requires you to say a few hundred to a few thousand utterances. The process goes something like this.
UNIDENTIFIED SURROGATE: Things happen in pairs. I love to sleep. The sky is blue without clouds.
PATEL: Now she's going to go on like this for about three to four hours. The idea is to cover all the different combinations of the sounds that occur in the language. From this voice bank, we can now say any new utterance like, I love chocolate - everyone needs to be able to say that - fish through that database and find all the segments necessary to say that utterance.
WOMAN: I love chocolate.
PATEL: What happens next is best described by my daughter's analogy. She's 6. She calls it mixing colors to paint voices. It's exactly that. Samantha's voice is like a concentrated sample of red food dye, which we can infuse into the recordings of her surrogate to get a pink voice.
PATEL: So now Samantha can say this...
SAMANTHA: This voice sounds like the real me. I helped make my new voice. I can't wait to use my new voice with my friends.
RAZ: That's incredible.
PATEL: I'm so glad you think that's incredible because, you know, as a scientist, though - I mean, there's been so many versions of that process. And even if they're not perfect, even if they have crackliness to the voice, even if they're not absolutely perfect in terms of speech quality, they're made with a little bit of them. And that is such a powerful thing. So for Samantha, when she heard her voice, there's this smile that spread across her face when she heard it for the first time. What I love about that is it's not her jumping up and down and saying, you know, yeah, this is my voice. It's this sort of slow realization that this is who I am.
RAZ: Speech scientist Rupal Patel. In a moment, we'll hear more about her research and another young girl who just got a brand-new non-Stephen Hawking-like voice. I'm Guy Raz. Stay with us. You're listening to the TED Radio Hour from NPR.
RAZ: It's the TED Radio Hour from NPR. I'm Guy Raz. And on the show today, extrasensory - technology and behavior that can actually stretch our senses. And we were just talking with speech scientist Rupal Patel who designs devices for people who can't speak because she wants them to sound more like humans, like themselves rather than robots.
PATEL: Many times, people who have these communication devices don't use their devices that often. And that's, you know, that's terrible because it kind of closes the world for them. I hope that this technology opens their door for them.
SHANNON WARD: My name is Shannon.
RAZ: And how old are you?
S. WARD: I'm 13 years old.
RAZ: This is Shannon Ward, and she was born with cerebral palsy. And a few months ago, Rupal essentially gave Shannon her own voice, and she built it on the vowel-like sounds that Shannon is able to make.
S. WARD: Yes. I think everyone deserves to be heard.
RAZ: So, Shannon, like, the voice you had before the one you have now, did you like it?
S. WARD: Not so much.
RAZ: So tell me about your new voice.
S. WARD: Totally love it.
RAZ: Does it feel like it's you?
S. WARD: That's a big fat yes.
JANINE WARD: To hear it, I was just stunned because I can hear my daughter, especially in certain words.
RAZ: This is Shannon's mom, Janine.
J. WARD: And when she says mom, in particular, I hear her. And that is a wonderful thing for any parent to experience.
RAZ: Does she seem like it's changed her?
J. WARD: Absolutely. It just increased her confidence, it increased her desire to want to use her device. I know, personally, that she loves to talk with her friends more. She stands out, and I think people take her differently when they hear a voice that sounds like a 13-year-old girl as opposed to a voice that sounds like a robot of an adult.
RAZ: OK, you have a 13-year-old daughter. She is now a teenager. If she's talking more, is she arguing more with you?
J. WARD: Absolutely. I know it might sound so bizarre, but there are times I'm like, can we unplug the device for a little bit because I have a kid that does not stop.
RAZ: So back to Rupal Patel, the speech scientist who started all of this - she's trying to get people all over the world to become voice donors to help people like Shannon.
PATEL: Yeah, and I think the beautiful thing about this is that as someone grows, what we'd like to see is that many people who use augmentative communication devices, these communication aids, use the same voice for much of their life. But you and I don't have the same voice today as we had a decade ago and a decade before that, certainly. It's layers and layers of experiences and changes in our body that change our voice. That should be afforded to these individuals as well.
(SOUNDBITE OF TED TALK)
PATEL: What I want to share with you next is how I envision taking this work to that next level. I imagine a whole world of surrogate donors from all walks of life, different sizes, different ages coming together in this voice drive to give people voices that are as colorful as their personalities. About five years ago, we built our very first voice for a little boy named William. When his mom first heard this voice, she said this is what William would have sounded like had he been able to speak. And then I saw William typing a message on his device. I wondered what was he thinking? Imagine carrying around someone else's voice for nine years and finally finding your own voice. Imagine that. This is what Williams said. Never heard me before. Thank you.
RAZ: Rupal Patel is a speech scientist at Northeastern University. She and her collaborator Tim Bunnell started a project where almost anyone can donate their voice to people like Shannon and Samantha. And you can do it, too. If you want to help out, go to Vocalid.org and give them some of your voice. And of course, you can watch Rupal's full talk at TED.com.
NPR transcripts are created on a rush deadline by an NPR contractor. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.