Let’s break down what #ElevenLabs has made involving the estates of these deceased #IconicVoices, so that they can “read” your favorite books, articles, etc.
For those in #VFX, you should consider: if ElevenLabs has the estate agreeing to a voice model, you can bet your 🫏 that OpenAI is negotiating the LIKENESS for #Sora. It’s a killer use case, and insurance will be all over this like white on rice; as will #sagaftra and #iatse.
Now you have a magic box that can digitally resurrect the dead, on command, to say and do whatever you want (within the limitations of the agreement, for now), which can then be modified into whatever medium (see below, for #Volkswagen example from 20+ years ago).
OpenAI will demand a percentage premium of that, not some per-hour baseline charge, especially if they have exclusivity and work to make specific models as perfect of a simulacrum as they can. Throw in hours of motion capture from video, to gather performance nuances, gait, expressions, etc. The sky’s the limit when it comes to a simulation.
What happens when we can perfect something that, 30 - 50 years from now, the human upon which one is based, conflates a sim with themselves?
Did that give you chills? Imagine when present-day actors, who already bank their likeness, then license their likeness, so they can do commercials without breaking a contract or filming schedule (in another language, no less, in whatever hair, makeup, and costuming you can dream of for them).
I can appreciate having who many consider one of the best actors of the last century read Shakespeare, as a selling point for their #GenAI #SaaS.
However, ask any actor, or anyone who genuinely cares for the Arts, respond to the “reading” of #Shakespeare by such a thing. People think I’m a harsh critic?
It’s going to be bereft of any feeling, because the AI isn’t going to understand what to emphasize and what not to. For example, the updated voices for #ChatGPT, from #OpenAI, also use Eleven Labs.
I deeply question whether the AI understands what it’s reading with any more surface comprehension as it would give to a grocery list. Indeed, that’s part of the current challenge with these AI models, and much of the over-hyped, over-reaching discussions on #xrisk.
The voice model could just be a stochastic model that sounds fine for everyday conversation, but it’s *not* built for #IambicPentameter, because who speaks in metered poetry all day long?
For Olivier, they likely have trained it for this specific consideration. Having used ElevenLabs myself, I know it’s possible to remap a voice with another model, so Judy Garland could deliver Shakespeare like Olivier.
Here’s a 12m masterclass clip of #IanMcKellan dissecting Shakespeare so brilliantly, I struggle to consider a better example: https://lnkd.in/gN9Sn56F
AI is not reflecting on Shakespeare the way McKellan is, not presently anyway.
Volkswagen’s digital resurrection of Gene Kelly: https://lnkd.in/gzWB4N3s
Introducing ElevenLabs Iconic Voices.
Listen to your favorite books and articles voiced by Judy Garland, James Dean, Burt Reynolds and Sir Laurence Olivier on our Reader App. The app is completely free to download and use. Get it here: https://lnkd.in/egiC-ZhP
We’re thrilled to partner with their estates to make the voices of these legendary stars available to you as we advance our mission of making content accessible in any language and voice. Read the full announcement here: https://lnkd.in/eHp37aUa