AI Debate Stirs in U.K. as Actress Replaced on BBC Project by Artificially Generated Voice

[highplainsdem]

Source: Rolling Stone

The AI battle is heating up in the U.K. after British actress Sara Poyzer was dropped from an upcoming BBC project in favor of an artificially generated voice.

Poyzer, who has starred in the musical staging of Mamma Mia! for a decade, shared a screenshot of the email from an unspecified BBC production member alerting her that her services on the unspecified project were no longer needed as the use of AI was permitted.

“Sorry for the delay,” the email reads. “We have had the approval from the BBC to use the AI generated voice so we won’t need Sara anymore.” Poyzer captioned the screenshot, “Sobering.”

-snip-

As the Hollywood Reporter notes, in 2022, the BBC actually backed a campaign by the British actors union Equity called “Stop AI Stealing the Show.”

-snip-

Read more: https://www.rollingstone.com/tv-movies/tv-movie-news/ai-debate-actress-dropped-bbc-project-artificially-generated-voice-1234995786/

---

Link to tweet

[intrepidity]

1. I want an app that does this

I want to listen to audiobooks and podcasts in voices other than the originals. There are just so many voices that irritate me and cause me to not listen to content that I would otherwise like to consume.. I want to choose the accent, too.

So, surely by now there is an app that will replace audio on-the-fly with custom options, right?

[highplainsdem]

2. If what you want is an AI app to give you "custom options"... then I wonder how you'd like such an AI app

to be trained?

From what I've read, the most common datasets are stolen recordings the AI company had no legal right to use and/or recordings made by very poor people, often in Third World countries, who have good voices and are desperate enough to sign away rights to use their voices in perpetuity, in any way, for just a few dollars.

Personally, I feel that anyone who wants a book read by AI and doesn't care if the AI used hurts actors' livelihoods and exploits humans should get an audiobook that sounds like AI rather than a human voice, since their overriding "custom option" was choosing AI.

But you could always ask people you know with voices and accents you like to record themselves reading books for you, and pay them what they think that's worth.

[intrepidity]

5. So, my desire for such a feature

(which will soon be technically ubiquitous) is separate from the issues you raise.

All that has to happen is for *one* person with the voice and accent that is pleasing to my ear, to sell/donate their reading of (whatever specs the tech needs) to train an AI voice to do on-the-fly translation.

I doubt it will be very complicated in 5 years, if not sooner.

Will it put certain people out of business? Like *all* new tech does, yes. And? That's news?

[AnrothElf]

3. YOu're kidding... right?

... Right?

[intrepidity]

4. Why?

Yes, I'm very serious. I spend many hours every day listening to various content through voices. I'm astonished at the number of people who record their voice for consumption but who lack good speaking skills, and it interferes with my concentration. Vocal fry is one of the worst, but by no means the only one. It is truly a big issue for me. There are several podcasts that I have unsubscribed from but whose content I *really* would like access to.

There's a reason for those "sample" buttons you can try before acquiring an audiobook. If the reason is different than what I've described, perhaps you can enlighten me?