Character.ai, a chatbot web application, has the following warning that I'm sure may apply here.
"Remember: Everything characters say is made up!"
-
Get involved.
We want your input!
Apply for Membership and join the conversations about everything related to broadcasting.
After we receive your registration, a moderator will review it. After your registration is approved, you will be permitted to post.
If you use a disposable or false email address, your registration will be rejected.
After your membership is approved, please take a minute to tell us a little bit about yourself.
https://www.radiodiscussions.com/forums/introduce-yourself.1088/
Thanks in advance and have fun!
RadioDiscussions Administrators
RadioGPT, AI hosts and the future of human on-air talent
- Thread starter Theater of My Mind
- Start date
"Remember: Everything characters say is made up!"
[song plays] You're listening to the NEW 105.7!!! More of your favorites from the 80's, 90's and beyond!!!" [song plays] "That was (insert name of song or artist here)." [song plays] "Stay tuned for more of your favorites from the 80's, 90's and beyond!! on the NEW 105.7!!! [5 commercials in a row] [song plays] etc. etc. ad infinitum. Which is typical on a lot of music stations, especially during times that aren't morning or afternoon drive.
Does it seem as if it would be that hard to program?
And don't forget the potential use on radio PSA's and commercials.
Ditto with Jamie Dupree, former Cox radio Washington reporter
If you're running a small station in Podunk, Wyoming, but still wants to sound as "local" as possible, programming a voice bot instead of asking VTing air talent to cut multiple extra liners has to be seen as a plus.
Sure...let's have a look:
Well, lot's of big-dogs of the industry there on that list, it would seem.
I would surmise that a large percentage of the general radio-listening population (e.g. NOT radio insiders or radio geeks...) wouldn't have a clue if a daypart was being done by an AI-driven computer voice.
As i mentioned in a prior comment in this thread, someone will be first. Many others will look at that cost-benefit analysis to see if it hurts or helps.
The National Weather Service has been doing this for 20 years on NOAA stations, including Spanish language voices. You can hear samples here:
With synthesized AI voices comes another possibility: AI voices that can mimic voices of past broadcasters. Talk about 'deep fakes.'
These are nothing like that. If you go back up and click the link for the demo, the RadioGPT voices (especially one of them) are much more natural.
Unlikely. You're not aiming at an audience that has great nostalgia for jocks of a different time. Or jocks, period.
As useful as this would be in small markets, I'm betting it's not going to be cheap early on. It may make more sense to continue VTing for the really small ones.
For the companies that can afford to license it, it's likely to be more about dumping benefit costs for full-time humans than anything else.
Everyone knows what it sounds like. It sounds like crap.
My bad, I listened to the AI demo after I posted, and like someone else mentioned, to people other than radio fans it may be hard to tell it's fake.
You'd think that they could get these to where they don't sound like squawk boxes by now, but the Spanish ones don't sound as bad as the English ones.
You're probably right, it's not going to be cheap. Yet, it could be revolutionary.
There's someone sitting in a C-suite level office that's formulating where this goes on-air first.
Version 2.0 is going to be taking the vocal characterizations of <someone known> and having the system replicate that voice in dozens of markets at the same time.
<someone known> will get a big check, and will never do voice work naturally again.
This could also extend to talent right now, like Ryan Seacrest, and other syndicated shows.
Thing about it is; you wouldn't dare have RadioGPT or TalkGPT present a newscast without significant intervention, mainly because GPT gets it's information from the public Internet. Between trolls, foreign bots, and erroneous Wikipedia posts, I'll bet an AI-generated newscast would be wrong 50% of the time.
Also EAS.
- Status