CADA's announcer, Thy, is just a bad cloned AI voice
There’s a show on Australian youth radio station CADA, 11am-3pm, that’s called Workdays with Thy. That’s Thy up there, in the image just above. CADA is on FM in Sydney, and on DAB in Australian cities nationwide; and, of course, on the iHeartRadio app. Thy has been broadcasting this show for the last six months.
Last week, a website called The Carpet wondered what was up, after being curious why Thy isn’t anywhere on social media, which the website says is quite odd for a 20 year-old woman in the media.
I’ve just noticed that the image on the CADA website is called beth-final-1-e1729730986847.jpg - and every other image of a person on that website seems to have their name in the image’s filename somehow. Is her name really Beth? ARN wouldn’t lie to us, would they?
Yes, they would. Thy isn’t really doing a show on CADA. Her voice has been cloned by ARN, fed into ElevenLabs, and she’s entirely AI-driven. Someone types a few things for her to say, and “she” says it. Apparently she works for ARN in their accounts department, though I’ve spent quite some time on LinkedIn and I can’t find anyone called Thy in accounts. Or Beth, for that matter. I think they’re lying again.
I listened and it made me sad
On working out how angry I should be about this, I went to try to listen to it. This was a mistake.
Having to listen to six telescoped hours of CADA has made me more despairing of today’s radio than I have any right to be. CADA is a radio station consisting of audio files played at random. Sometimes you get the news, in a jazzy format, at about six hours past the hour. Sometimes you don’t. Sometimes you get random ad breaks full of promotions for podcasts. There’s no reason to keep listening, no feeling of community, just automated music back to back. This is a waste of an FM frequency, produced by people who don’t care. It’s dreadful, dire, stuff.
Anyway, here are three breaks from “Thy”. You’ll not be able to hear much of what she says, since her voice is too quiet in the mix, and it’s obvious that nobody at ARN has even bothered to listen to it before it aired. She appears to only do three presenter breaks in this two hour segment. She’s not really replaced anyone as a radio announcer on this station - she says nothing of interest, and it’s just poorly-written shit.
This website says that ARN has a deal with ElevenLabs, which is a voice synthesis tool. But this clone is nothing to be proud about. I’m surprised ARN deems it acceptable for use on-air, but then, of everything I’ve heard of this radio station already, it probably doesn’t sound any worse than the awfulness of what it surrounds. Of what you can hear in the badly-mixed output, it’s a bad, bad clone.
For what it’s worth, my own cloned voice is over here - hit the listen button on this website page - doing a read that I think is not great, but palpably better than “Thy”. And, I bothered to mix it properly, because - unlike ARN - I care about the output, so you can actually hear it. Seriously, what the hell is CADA’s audio supposed to be? Is anyone going home from ARN thinking “I’ve produced some good radio today”, having made that?
Honesty is the best policy
Anyway, rants about the quality of the dismal output aside, this breaks the #1 rule of not lying to your audience. If the future of radio is human connection and shared experience - which I believe it is - then at the very least you should be honest with your audience and say if your apparent main on-air talent is actually AI.
Surely - surely - CADA’s audience (if there is one) is accepting of AI - and “A.I. Thy” could have a little fun doing a real radio show. “A.I. Thy with you on CADA… what’s up with Donald Trump right now? Anyone know? Have they tried turning him off and on? It’s what they do with me! Hey, here’s a throwback from Warren G…” - that would be fun, and if it was mixed by someone who actually had fucking ears, it would be quite a good listen.
Had ARN actually been honest and straight-up with its audience, that “Thy” was AI, then I reckon it would be absolutely fine. But, as it is, it’s the Milli Vanilli of Sydney radio - caught out lying to its audience, peddling a badly cloned voice as a real radio personality. ARN fucks up the one thing that radio has going for it - human connection and shared experience - by faking the very human connection it has set out to achieve.
This is depressing bullshit; but, no more depressing than ARN’s big-name stars Kyle and Jackie O broadcasting awful indecent material to children at breakfast on KIIS for over a year, though you have to hand it to them, at least someone’s actually producing that output, rather than the CADA shuffle-o-matic via Winamp.
Awful, awful crap. If ARN are in charge of the future of radio, let’s all pack up and go home. Irredeemable. Get in the sea, and leave radio to people who care.
Anyway, I’d best think about how to make my views clear about this, because I’m worried people won’t understand what I think.