>“A good actor does more than just read the story. They imbue it with core emotions and bring out the subtext, elevating the written words with empathy and nuance,” Rachel says.
So true!
“We bring the human lived experience to each story in a way that cannot be replicated.”
Seems like maybe this is specifically not the case. In any case they should be very well compensated if you are going to use their data to eliminate their future as audio book readers. I hate that Apple which could afford to do this essentially steals from them.
>“The voice is how voice actors make a living,” Friedlander added, “and this is literally taking the words out of our mouths without our consent.”
The way Apple and Spotify went about this in this circumstance is bad. They should face charges if they do these things without consent.
That being said, while I feel for those who make money from their voice in audiobooks, I can't help but feel like that line of work won't last long into the future. This isn't podcasts we're talking about, it's audiobooks. They're reading a script and once those algs can reproduce inflections and personality of voice perfectly I don't see how anyone could expect to make money doing this except the very few recognizable voices out there. Even then, I'm sure those celebrities would rather just get payed by Apple and the like to use their AI-voice so they don't have to actually sit there for hours reading a book.
Also, I think AI-generated audiobooks are good overall. It will make audiobooks for every single book out there within reach and super cheap for those who would otherwise need to pay to create one. (recording studio, time & labor).
I don't think this is true at all. Maybe I'm vastly underestimating how good these AIs are or soon will be.
I recently listened to the Vorkosigan saga, ~15 books all narrated by one person (Grover Gardner). Across these books there are probably 100 different voiced characters, and some of them show up in one book and then don't show back up until 5 or so books later. They each have their own voice and the voice is consistent throughout the whole series.
Is this something you think AI is capable of, or will soon be capable of?
All of The Expanse audiobooks were narrated by Jefferson Mays, except for the third book where they got some other guy to do it. Fans were so upset by this that they basically forced the company making to audiobooks to re-record the whole book a second time with Jefferson Mays.
I don’t think an AI is going to pull this off. Maybe with human help to tell it which characters should use which voices? Even then I think there’s a lot of “acting” in voice acting that an AI would fail miserably at. It might get 95% if it right, but I think the 5% would be off putting enough to make up for it.
I think in a case like that, it would be hard for AI to get it done right any time soon. But I think that case is kind of an outlier in the audiobooks world. Or potentially I just listen to a lot of non-fiction..
> VALL-E emerges in-context learning capabilities and can be used to synthesize high-quality personalized speech with only a 3-second enrolled recording of an unseen speaker as an acoustic prompt.
> She was shocked to find a clause titled “Machine Learning” near the bottom of her lengthy agreement with Findaway
I have mixed feelings about this. If you're signing a contract for work you're doing, it should be expected that you've read and understood what it is you're agreeing to. The excuse of "It was buried in with other clauses" doesn't seem terribly convincing. This just seems like the panic du-jour.
How do you know if you truly understand something?
I find it highly unlikely that you have correctly interpreted every single clause of every single document that you've ever signed for work, and I find it even less likely that your interpretation of these clauses would match how a court would interpret them if it ever went to court.
from your response it appeara thst you have misunderstood my comment, which leads me to believe that you frequently misunderstand contracts that you sign.
But take a look at the reading level of a ToS, now compare that to the average reading level of a person who signs that contract.
There is a vast gap between those two numbers, and that gap is intentional.
When two parties meet to sign a contract and that contract is entirely crafted by one party, the only party with unfathomably large resources and that contract is so hideously one sided in all facets down to the reading level it isn't hard to see that this one party is behaving unethically.
Apple is the 40 something dude with a bit of money and life experiences who uses these things to court exclusively emotionally vulnerable 18-22 year olds. It's not illegal it's just scummy as fuck and they should be judged for it.
I agree, most ToS are essentially nonsense to most people.
This is also not a contract between the narrator and apple, it's the narrator and their distributor. They should absolutely be reviewing that thoroughly, and ideally with their own counsel.
So the audio book narrators have two bad options: A. refuse to let their voice be used to train ML I.e don’t work because I’m sure all contracts have or will soon have that clause, or B. Work for a few years training their robot replacements and then lose work due to being replaced by robots.
Author here. These are not "work contracts" the way you may be thinking. When we sign with publishers etc we do have agents and lawyers. This was basically in the terms of service and was slipped in quietly after not including it for years. It would be like Photoshop quietly adding a clause that they had the rights to use your art in AI.
Should every digital artist read the TOS... again and again? Every year? Every month? It's a nice idea to be that informed but not realistic especially with the number of websites we use to do our work.
Additionally a lot of times it's the authors who upload not the narrator so technically the narrator did not even agree to even that much.
21 comments
[ 3.0 ms ] story [ 58.6 ms ] threadSo true!
“We bring the human lived experience to each story in a way that cannot be replicated.”
Seems like maybe this is specifically not the case. In any case they should be very well compensated if you are going to use their data to eliminate their future as audio book readers. I hate that Apple which could afford to do this essentially steals from them.
The way Apple and Spotify went about this in this circumstance is bad. They should face charges if they do these things without consent.
That being said, while I feel for those who make money from their voice in audiobooks, I can't help but feel like that line of work won't last long into the future. This isn't podcasts we're talking about, it's audiobooks. They're reading a script and once those algs can reproduce inflections and personality of voice perfectly I don't see how anyone could expect to make money doing this except the very few recognizable voices out there. Even then, I'm sure those celebrities would rather just get payed by Apple and the like to use their AI-voice so they don't have to actually sit there for hours reading a book.
Also, I think AI-generated audiobooks are good overall. It will make audiobooks for every single book out there within reach and super cheap for those who would otherwise need to pay to create one. (recording studio, time & labor).
I recently listened to the Vorkosigan saga, ~15 books all narrated by one person (Grover Gardner). Across these books there are probably 100 different voiced characters, and some of them show up in one book and then don't show back up until 5 or so books later. They each have their own voice and the voice is consistent throughout the whole series.
Is this something you think AI is capable of, or will soon be capable of?
But I think you're overestimating the quality of the "average" audiobook and its narration
A sizeable swarth of books don't lend themselves to such finessed narration
I don’t think an AI is going to pull this off. Maybe with human help to tell it which characters should use which voices? Even then I think there’s a lot of “acting” in voice acting that an AI would fail miserably at. It might get 95% if it right, but I think the 5% would be off putting enough to make up for it.
> VALL-E emerges in-context learning capabilities and can be used to synthesize high-quality personalized speech with only a 3-second enrolled recording of an unseen speaker as an acoustic prompt.
I have mixed feelings about this. If you're signing a contract for work you're doing, it should be expected that you've read and understood what it is you're agreeing to. The excuse of "It was buried in with other clauses" doesn't seem terribly convincing. This just seems like the panic du-jour.
If I don't understand something I ask about it, or talk to the other party to have it clarified, re-worded, or removed.
I find it highly unlikely that you have correctly interpreted every single clause of every single document that you've ever signed for work, and I find it even less likely that your interpretation of these clauses would match how a court would interpret them if it ever went to court.
Nothing in the article seems to imply anyone misinterpreted their contract. They didn't read it.
But take a look at the reading level of a ToS, now compare that to the average reading level of a person who signs that contract.
There is a vast gap between those two numbers, and that gap is intentional.
When two parties meet to sign a contract and that contract is entirely crafted by one party, the only party with unfathomably large resources and that contract is so hideously one sided in all facets down to the reading level it isn't hard to see that this one party is behaving unethically.
Apple is the 40 something dude with a bit of money and life experiences who uses these things to court exclusively emotionally vulnerable 18-22 year olds. It's not illegal it's just scummy as fuck and they should be judged for it.
This is also not a contract between the narrator and apple, it's the narrator and their distributor. They should absolutely be reviewing that thoroughly, and ideally with their own counsel.
Should every digital artist read the TOS... again and again? Every year? Every month? It's a nice idea to be that informed but not realistic especially with the number of websites we use to do our work.
Additionally a lot of times it's the authors who upload not the narrator so technically the narrator did not even agree to even that much.
They literally did this. I fully sympathise with you, not sure why OP thinks contracts are always simple and innocent.
https://m.dpreview.com/news/6341509927/adobes-content-analys...