There are more than 10 million people in the world who are unable to speak. Like the young girl in the video above, they rely on computerized devices to provide their voices. But because there are so few recorded voices available, the same voice used by theoretical physicist and author Stephen Hawking is used by little children.
When speech scientist Rupal Patel was at a conference and saw a young girl talking to a grown man and both were using the same synthetic computer voice, she knew something had to be done. There were hundreds of people at the conference who were unable to speak and they were using artificial voices that didn't fit their personalities or their bodies.
The generic voice and lack of individualization really struck a chord, so Patel worked to spearhead the funding and technology to create custom voices for custom people, as she explains in the video below. The company is called VocaliD.
The way it works
To start the process, a voice donor records a series of short stories and sentences. They don't record every single word a person might say, but they typically cover all the different combinations of sounds that occur in language. The process may take between five and seven hours, but it doesn't have to be done all at once and it can be done in the contributor's home.
The voice donor can stop and start, picking up whenever it's convenient. They're encouraged to record from the same place each time so the sound is consistent.
Those recordings are then divided into little snippets of speech to populate a database.
The person who will receive the voice also records a few distinctive sounds. The VocaliD team then searches the voice database for a perfect vocal match. That donated voice is infused with the donor's sounds. Combined, this makes a unique voice.
Here, 4-year-old Oscar listens to his new voice for the first time.
The Human Voicebank
So far, more than 14,000 speakers from more than 110 countries have contributed to what the company calls "The Human Voicebank." People who want to donate can record from anywhere they have a computer or phone, sharing their voice with anyone who needs it or even banking it for themselves.
"English speakers from all over the world share their voice as part of our Human Voicebank initiative because they want to help bring speech to the speechless," VocaliD spokesperson Elisabeth Nuboer tells MNN. "We have voice contributors from over 120 countries ranging in age from 6 to 91 and welcome everyone and all their wonderful accents."
Watch what happens when one young girl helps give her sister a voice.