As technology continues to advance, we find ourselves with more and more voice-activated devices like smartwatches and microwaves. Some folks speak with their hands, though, making items like this sign language ring so important. And some who lack a voice, like the famous Stephen Hawking, use a modulated artificial voice that is impersonal at best and robotic at worst. Rupal Patel and Tim Bunnell, speech scientists at Northeastern University, think that’s unfair.
Patel and Bunnell want to offer a voice to those who can’t speak. They’re creating custom-made voices that focus on everything from pitch and volume to pronunciation of certain letters like “ch.” They do so with the help of voice donors, a concept right out of science fiction. They find someone from a similar demographic in all regards to the one requiring a voice: age, gender, size, what part of the country s/he’s from, and so forth. That person then spends several hours recording more than 3,000 different phrases that capture every potential instance of the human language. With each phrase, the new simulated voice becomes better. Special software takes all those phrases, runs them through the electronic ringer, and reverse-engineers a voice to fit the person needing one to replace the robotic voice that synthesizers of today are burdened with.
Of course, there are a lot of people who need voices, and the process takes a significant amount of time for donors. The incentive is helping your fellow man, but most people probably don’t even realize this is a path to doing so. So Patel and Bunnel began VocaliD.org, which allows both potential donors and those seek a custom voice to sign up to offer/receive its services. More than 10,000 people have signed up to be voice donors, and several hundred have signed up to receive new voices. Patel’s happy enough with the numbers, but she wants to hit the one million donated-voice mark by 2020.