I don’t know, the “Spanish prisoner” is a scam that seems to be reinvented every few years every time we see a little bit of a change in technology. It wouldn’t take much to fake a person’s voice with a trained model, especially if that person has an online profile open to the public where they post content in their own voice.
I could easily conceive some tricks to get clips of a person’s voice without them realizing. I’d write them out but… that would be stupid of me. Humans have more vulnerabilities than computers.
Yeah, unless this person runs a YouTube or podcast it seems implausible. What would you train a random AI on for the normal person?
I could see a situation where you hack a phone, get the contacts and call history, pick the 1st or 2nd most dialed number, have a bot call that person to get samples, then go back to the original phone and try this… I mean, eventually you’d get a hit?
Not sure I buy this tbh.
I don’t know, the “Spanish prisoner” is a scam that seems to be reinvented every few years every time we see a little bit of a change in technology. It wouldn’t take much to fake a person’s voice with a trained model, especially if that person has an online profile open to the public where they post content in their own voice.
The Spanish Prisoner is also an excellent David Mamet film.
A: The Spanish Prisoner is also an excellent…
B: …excellent
A: David… David Mamet.
B: What I’m saying is, when it comes to excellent films
A: That’s what I’m saying
B: (simultaneously) David Mamet
A: (simultaneously) Mamet
B: it’s excellent
A: hm. The best.
Yeah, it has some sus vibes. I’m usually far too trusting, but here even my bullshit detectors rang
You know that old adage “Never attribute to malice that which can be easily explained by stupidity”?
We need a new one along the lines of “Never attribute to truth that which can be easily explained by attention-starved teenagers”
I could easily conceive some tricks to get clips of a person’s voice without them realizing. I’d write them out but… that would be stupid of me. Humans have more vulnerabilities than computers.
Yeah, unless this person runs a YouTube or podcast it seems implausible. What would you train a random AI on for the normal person?
I could see a situation where you hack a phone, get the contacts and call history, pick the 1st or 2nd most dialed number, have a bot call that person to get samples, then go back to the original phone and try this… I mean, eventually you’d get a hit?
deleted by creator