Rendered at 20:07:15 GMT+0000 (Coordinated Universal Time) with Cloudflare Workers.
tanseydavid 2 days ago [-]
I wonder how John Fogerty feels about this? He was sued (unsuccessfully) for sounding too much like himself.
MisterTea 2 days ago [-]
What is the future for the voice actors then? Once the AI is trained they don't need to hire the kid for further recording sessions so they can cut them loose, use the AI and pay whatever royalties.
At some point they wont even need people and you can just ask for "British Cockney spoken by an 8 year old female with a slight Chinese accent" then fine tune the results and paste in the dialogue.
ronsor 2 days ago [-]
> "British Cockney spoken by an 8 year old female with a slight Chinese accent"
This is already the case. The voice acting industry has inertia, and many union contracts have minimum hiring quotas.
In practice, if you consider the output good enough, existing models can generate arbitrary voices that don't correspond to any person. No one has to sign or license anything.
riddley 2 days ago [-]
I'm a knowledge worker who, like many, have had a stressful time with the rise of LLMs. The other day I realized that we've had drum machines since the late 70s but we still have drummers. I'm sure there are flaws in this analogy but it helped me to calm down.
rolph 2 days ago [-]
what is the future of having your own voice inaccessable as a chattel of labour due to contract law?
add-sub-mul-div 2 days ago [-]
What's the future of any profession that can now be sloppily automated? Not great.
kazinator 2 days ago [-]
That's like a contract for milk delivery where the customer adds a clause that all the cows are belong to them.
At some point they wont even need people and you can just ask for "British Cockney spoken by an 8 year old female with a slight Chinese accent" then fine tune the results and paste in the dialogue.
This is already the case. The voice acting industry has inertia, and many union contracts have minimum hiring quotas.
In practice, if you consider the output good enough, existing models can generate arbitrary voices that don't correspond to any person. No one has to sign or license anything.