Want To Be A Siri-Like Voice Surrogate For The Speech Impaired? You Can With VocaliD
There are millions of people with speech impairments, and currently the vast majority of them must use generic synthesized voices to communicate. (Think Stephen Hawking.) Apple’s Siri is a voice built from human samples, although in her case the voice wasn’t “mixed” with anyone else’s.
However, the aforementioned professionals have a company called VocaliD that is working to develop a way for patients with speech impairments to be matched with voice donors that fit their physiology (eg, male, female, tall,) as well as their origin (eg, from the South, from New York) and vocal qualities (eg, raspy, low, breathy).
They record whatever vocal sounds the patient can offer, match them with a donor voice, and then synthesize them together to create a voice that sounds as much like what the patient’s voice should sound like as possible. Thus, if you were a donor, you wouldn’t someday overhear your own voice ordering food at the next table over at a restaurant; it would sound like a different voice, but one that ideally suited the person using it.
Further, as VocaliD points out on its website, if you bank your own voice, you could recreate it later on should you ever find yourself needing it.
You can get a simple primer about how the technology works, complete with audio samples, here. (Spoiler: It’s beautifully brief, simple, and amazing.)
The “donation” currently requires donors to read certains for a total of about 2-3 hours. However, as mentioned above, VocaliD wants to develop an iPhone app that would let you record your voice into your mobile device on your own time and in that way build a massive database of different voices. The team wants serviceable samples of essentially any and every type and quality of voice that there is, including children’s voices.
They’re calling it the Human Voicebank Initiative, and if you want to donate time, money, or your voice, you can learn more and sign up at VocaliD.org.