A groundbreaking center has been opened in the US aimed at preserving and restoring the voices of artificial intelligence. Researchers hope it will change the lives of people who lose their ability to speak.
A collaborative effort by Northeastern University in Boston with VocaliD could play an important role in preserving the sense of identity of patients with conditions such as throat cancer or motor neurone disease by giving them the opportunity to re-imagine self-knowledge. generated language has become impossible.
The first of its kind is Prof. Rupal Patel, the founder and CEO of VocaliD, as senior researcher of the center.
While Patel said the company has already given individuals the opportunity to keep track of their voices in their own homes, many people in reality either lack high-quality recording equipment or make recordings with background noise. Patel said that it is also necessary to offer greater support to those in need of such services and to ensure that patients are aware of their availability in good time.
"Often they will come to us at the last minute," she said. "They do not have enough time to narrow their voices, and they're just so entangled in their illness and surgery ̵
As a result, the company has teamed with Northeastern – where Patel is currently a faculty member on leave – to bring the technology into the community. The result is the Voice Preservation Clinic, a center where people who normally run the risk of losing their voice can record themselves as part of a "legacy" project.
The approach is far more sophisticated than dissecting words and then stringing them. Instead, they use machine learning algorithms – a kind of artificial intelligence. The first step is to "bank" the vote by offering poems, short stories or speeches on a range of topics that they record in a special booth.
"What we let them do is record about two to three hours of speaking time. From these records, we can then create an AI-generated voice engine that essentially sounds like it, "said Patel. In fact, it can say words in the user's voice that have never been recorded.
If the digital voice is installed in the associated app, whether it's a phone or a special device, the user can enter what they want to say. Audio generates the sentences in the user's voice.
"With the cancer population, they have control of their hands and they can communicate – but they want to communicate like themselves," Patel said.
The Legacy Project, she added that it differs from the company's other services, where employees can donate footage of themselves to create a "tailor-made" voice for those who can not speak ,
Technology, Patel says, is improving rapidly. Voices sound more and more human, and while there are still some ways to change voices' voices, the team has begun to develop filters that give users more choice in how phrases are expressed.
Patel said the team could also age a person's voice so that it grows, but only to a point. It is not yet possible to turn a child's voice into that of a teenager or an adult, though the team can turn them into the voice of an older child.
The recording service is free, but costs 1,499 USD (1,360 GBP) to build a vote. The clinic said, however, that they would not charge customer fees in the first year.
Patel said that anyone can come and classify their voice, but give priority to those who expect to lose their ability to speak.