Neal Stephenson's "REAMDE" featured a game which did exactly that. That was published in 2011.

Now in 2024, after the boom of generative AI, and audio deepfakes becoming commonplace, this could also be considered "bleedin' obvious" and thus not patentable

I suspect prior art may already exist. Text-to-speech systems have existed for decades and voice synthesis has been around for a few years. Is this another case of an American company trying to patent a pipeline? The only slightly novel thing I can think of is maybe getting Einstein to speak Klingon correctly, though I wonder if the all the voice shifts are possible? I know that my voice changes depending on which language I'm speaking.

if it were used in elections.

The tape recorder and video camera gave us the "Age of Accountability" where incontrovertible evidence could be produced to expose liars, cheats and criminals

Now, generative AI gives us the Age of Plausible Deniability, where anyone can simply claim that whatever evidence it is must be "fake news", especially if they are powerful enough to be widely believed

It's worse: anything not witnessed in person, face to face, with someone whom you personally know and can identify, cannot be trusted to have said/done what it claimed.

Already has. They tracked the GQP russian cult scammer that imitated President Biden's voice in Texas.

Or in a telephone fraud coming to your neighbourhood soon…

This patent application is very odd to a UK observer. Nowhere is any finite mechanism described that fulfils any of the proposed functionality, so it's an entirely conceptual document. If this high level kind of application can succeed in the US it means that entire disciplines could be locked down by opportunists without the need to actually implement any demonstrable invention. I do hope that's not the case, as it would stifle innovation, the very thing patents were created to encourage.

"it means that entire disciplines could be locked down by opportunists without the need to actually implement any demonstrable invention"

Yup - that's exactly how the US patent system works. Patents are only tested for originality or practicability when they are challenged by either a competitor, ie a holder of a similar patent, or when the holder uses said patent to crush (read: overwhelm and buyout) an "imposter" who actually gets something relatively similar to work.

From a UK perspective it is indeed a very, very odd system: only of any benefit to lawyers (guess who wrote the rules...)

Look, that old world we knew...It's gone. It doesn't exist. All that stuff we learned our whole lives. "Just because you can doesn't mean you should", wrong... they'll do it if it can be done. "Anything worth doing should be done right", wrong, it's all just instantly reproducible, objectified digital content now. "Two wrongs don't make a right", nope... it's eye for an eye now. Why you ask?

Because, you buy it. Someone will. These are companies. Their job is to sell you ish. These are not good people doing good things.

They will sell you ish in a bag if you will buy it for a penny, and someone will. Someone will spend a penny on dog ish in a bag if it's marketed right.

It's all cheaply produced, easily reproducible, instantaneous dog ish in a bag. And, people buy it.

It's a method patent (often called process patents in other countries, such as the UK), rather than a machine patent, as you described. They crop up in chemistry-related topics a fair amount, to describe a particular production process.

https://en.wikipedia.org/wiki/Method_(patent)

Edited to make clear: It's an appication at this stage, not a granted patent!

...this technology being deployed remotely on your BlueTooth equipped toothbrushes (yes they are out there).

Stop putting words in my mouth, just shut the f*** up and clean my teeth!

I'd have thought that either replacing a video's original speech track with one in another language and/or replacing it with a different speech in the same language would both show a rather obvious lip sync mismatch, but there's no indication of whether the AI model is capable of recreating lip movements to match the revised audio.

It would be nice to know whether if it has that capability.

Think of all the places where telephone calls are recorded for training and quality purposes. Not stored securely and able to be listened to/downloaded by anyone.

My bank uses "My voice is my password"