- Amazon Alexa may soon be able to talk to you in custom sounds– Like your dead grandmother’s voice.
- It’s a feat of speech synthesis technology, which has been around for a while, but now has the endorsement of major corporations.
- However, this vocal eternity comes with some risks, from bank fraud to putting words in the mouths of deceased influential figures.
In the very near future, Amazon’s popular voice assistant, Alexa, may look very different from the obedient (and impersonal) voice you’ve been accustomed to since its advent. 2014. In fact, the voice of your cloud-based digital assistant may be bouncing off your kitchen walls with the voice of your deceased grandmother, your wife, your best friend, or even Elvis Presley.
At least that’s what Rohit PrasadSenior Vice President of Amazon and Chief Scientist at Alexa, announce at Amazon’s re:MARS conference, a global artificial intelligence (AI) event hosted by Amazon founder and CEO Jeff Bezos over the summer. With just a minute-long audio sample, the technology can bring a loved one’s voice through the Echo device’s speakers.
use prasad short show To show the audience how new speech synthesis technology can help us form lasting memories of our deceased relatives. “Alexa, can my grandmother finish reading me The Wizard of Oz?asked a cute little boy speaking from Echo with big panda eyes. “Okay,” Alexa replied in her usual voice. Then the boy’s “grandmother” began telling the classic children’s novel. Prasad did not say exactly when this feature would be rolled out, nor were there any further details on how it would work.
More from Popular Mechanics
Robotic speech synthesizers have been around for a while, but they haven’t really made their way in pop culture Until the 1980s, when theoretical physicist Stephen Hawking began using it. To create a synthesized speech, you link together the parts of recorded speech that are stored in a database. “Amazon, specifically, is using a sound bank that they already have to build a base model. Then, they will adapt the base model accordingly,” Lee Malone Tells popular mechanics. Mallon is an app developer who works on projects for Alexa voice services and is the founder of good voicean app that keeps recorded stories that loved ones read aloud.
“Let’s say you speak English. They use the data of thousands and thousands or more of people who speak English as a basic type of language model, and then add your voice fingerprint to it, to generate your synthetic voice within a few minutes,” Malone explains. Your voiceprint is your real voice, with all its unique characteristics (think: voice biometrics).
ethical worms can
However, the fact that Amazon only takes a minute to reconstruct a person’s voice does not reflect a lifetime of feelings. Would a person be able to narrate a sentence in a state of terror or excitement and laughter at the same time? asks Malone. In other words, will the 60-second clip contain every inflection of a person’s voice? Malone believes that in the few successful cases where the synthesized sound is able to capture the exact feelings of the origin, the result can greatly help a person in processing grief.
However, in most cases, the end product can be disappointing, if not outlandish – at least until technology is advanced enough to blur the boundaries between real and synthetic sound. “We still have five to six years of indistinguishable from the real sound,” Malone says. Not to mention that speech synthesis in its current nascent state could open a large moral can of worms.
In February 2021, for example, Hollywood star Tom Cruise deep fake swept tik tok. Cruz showed off his CD collection and played Dave Matthews’ band song on guitar. There was a creep about fake media’s similarity to reality that worried many TikTok users: What if someone used a fake image (or audio) of us to act out an embarrassing scene and spread the synthetic media around the internet?
But things don’t get better in death either, because the deepfake technique might not allow us to rest in peace. In theory, anyone with access to our data — such as tweets, Facebook messages, voice notes, and emails — could resurrect our image through a deepfake, avatar, or chatbot without our ever consenting to such a thing when we were alive. . Creating an index from this data doesn’t always lead to organic or honest responses, said Irina Raiko, director of the Internet Ethics Program at the Markola Center for Applied Ethics at Santa Clara University. popular mechanics in 2021.
“If this is accepted, I think this could have a chilling effect on human communication,” Raiko says. “If I’m worried that anything I’m going to say might be used in a weird avatar of myself, I’m going to have to guess everything.”
Living people can discuss deepfakes and bring the perpetrators to court. But with the dead, especially those who died in the not so recent past (and those without active legal legacy), there is a greater chance of abuse. What would happen if Muhammad Ali spoke about racial tension in words he never said? The famous American boxer was a Muslim and a famous advocate for the rights of African Americans.
“Imagine what would happen if we took a voice on Ali now, with all these things With Salman Rushdieand puts words in his mouth – words he will never utter? ” Requests Rupal Patel, who is a professor in Northeastern University’s Department of Communication Sciences and Disorders and vice president of voice and accessibility at Veritone, a California-based artificial intelligence technology company. (Rushdie, a famous Indian-born British English writer, was stabbed in August before giving a lecture on the United States as a safe space for exiled writers.)
“We need to proactively prevent such egregious abuse, or else we could end up ‘misinterpreting the mark of a life-impressive personality,’” says Patel. Do it with other dead public figures, and you could end up tarnishing an entire legacy and dumping an already going community. In a state of imbalance.
Who really owns your voice?
With this new development, Amazon is promoting an existing technology, but we have not yet protected ourselves from the problems that might arise if audio technology spreads after its death.
“Your voice is your intellectual property,” Patel says. “There has to be some sort of control as to who can access the license for that voice, or who can control the voice engine once it’s been created, because there are huge risks otherwise… AI voice can be used to impersonate someone, and that might not Fools a human, but may deceive a voice authentication system like the ones used in banks. “Voice escort is a whole new class we don’t know how to deal with yet,” says Patel. popular mechanics.
And, as with every bit of exciting new technology championed by major companies, we may want to read the finer details first. to me machine learningAmazon employees listen to and rate the voice input on a regular basis. Amazon keeps a copy of everything Alexa records after you hear his name, and It saidAlexa eavesdrops on her masters regularly. “The algorithm can measure your age, your gender, whether English is your first or second language from subtle influences in your voice when you speak it, and much more,” Malone says.
that April 2022 Report From the University of Washington, UC Davis, UC Irvine, and Northeastern University found that Amazon shares Alexa data with 41 different ad partners. That’s probably the ultimate motive for getting Alexa to talk “from the other side,” Malone explains. “They do this to make it look more attractive and to keep Alexa alive, so you can keep going to your house.”
Stav Dimitropoulos’ scientific writing has appeared online or in print for BBC, Discover, Scientific American, Nature, Science, Runner’s World, The Daily Beast, and others. Stave put his sports and academic career on hold to become a journalist and get to know the world.
#Alexa #speak #voice #dead #grandmother