RadioGPT AI automated radio presenters

[LeeCavanagh]

Saw this, may interest you
From TechCrunch

Radio station gets part-time AI DJ based on its midday host
Lauren Forristal@laurenforristal / 11:16 PM GMT+1•June 14, 2023
Comment
Podcasting studio setup with recording equipment and microphone
Image Credits: Asia-Pacific Images Studio / Getty Images
Live 95.5, a radio station in Portland, Oregon, announced Tuesday that its midday host Ashley Elzinga will broadcast a cloned version of her voice — aka “AI Ashley” — to listeners every day from 10 a.m. to 3 p.m. They’ll be using Futuri Media’s “RadioGPT,” an AI-powered tool that uses GPT-4 to generate a script based on trending news and reads it with a synthetic voice.

The radio station noted that AI Ashley isn’t fully replacing “traditional Ashley,” which has been a huge concern for a lot of local DJs. Radio stations have cut down their broadcasts in recent years due to rising costs or in a shift to embrace tech like AI. In 2020, iHeartMedia restructured its organization, laid off hundreds of people nationwide and invested in artificial intelligence.

Live 95.5’s parent company, Alpha Media, assured us that Elzinga would not be losing her job and would still be paid the same salary.


“It’s a hybrid situation where we’ll have traditional Ashley on during some segments, and we’ll have AI Ashley on during other segments,” Phil Becker, Alpha Media EVP of Content, explained to TechCrunch. “In an instance where AI Ashley would be broadcasting, the traditional Ashley might be doing something in the community, managing social posts or working on digital assets or the other elements that come with the job.”

Becker also noted that Alpha Media isn’t using RadioGPT to save costs. It’s meant to be an efficient tool for radio hosts to have in their toolset.

Listeners may be wary of AI DJs, since they have tuned into local radio personalities for years and don’t necessarily want to listen to a robotic-sounding voice on their way to work in the morning. That’s why Live 95.5 trained the AI to sound like an existing radio host that the audience has already connected with.

As seen in a tweet where Elzinga shows off her AI DJ counterpart, the voice is similar to the real thing but very obvious that it’s AI. The synthetic voice also introduces itself as AI Ashley, so listeners know that an AI is speaking and not a real human.


“One of the absolute most important parts of this is that we’re transparent with the listener. It’s not our intent to ever deceive anybody,” added Becker.


Twitter users expressed their disappointment in the new decision. One user even said the AI voice was disrespectful to the entire radio profession.


While Live 95.5 is one of the first radio stations to use RadioGPT, it certainly won’t be the last. Alpha Media’s KUFO Freedom 970, another Portland-based radio station, will also be using the AI tech.

Alpha Media oversees more than 200 radio stations in the U.S. The company claims to be the first radio broadcasting company in the world to have an AI DJ.

Another Futuri partner is Rogers Sports & Media, which owns 55 stations and over 29 podcasts in Canada.


Image Credits: RadioGPT


When Futuri launched the beta version of RadioGPT in February, people questioned the ramifications of the AI tech. One major concern was if there was a risk that AI DJs could spread misinformation. Fact-checking isn’t ChatGPT’s strong suit and it has been known to “hallucinate,” meaning it confidently provides users with fake information.


Bing and Google both have AI chatbots that users have criticized for making factual errors.

ChatGPT relies on the help of human moderation, so it’s important that radio stations using RadioGPT have real people checking the generated content.

Futuri’s tech scans over 250,000 news sources as well as stories on Facebook, Twitter and Instagram to identify which topics are trending locally. RadioGPT will take these trending stories and then use GPT-3 to create a script. Before the AI voices relay the info to listeners in that local area, Becker claims that a team of human moderators reviews and edits the content before it goes out on the air.

“You’ll still have to have the original content creators touching it, checking it, proving it, all of those sorts of things,” Becker said.


Plus, Futuri CEO and founder Daniel Anstandig claims that safeguards are in place to prevent hallucinatory content from going out over the air.

“Our AI is programmed to cross-check and reduce the occurrence of speculative or inaccurate content, as is sometimes referred to as ‘hallucination,’” Anstandig claims. RadioGPT also “confirms that any generated content does not contain any offensive material,” he added

[famefm]

After hearing some of today's so called personality presenters AI would probably be an improvement not that's is saying much?

[yellowbeard]

I prefer the anarchy of a real presenter. However I would like it if it was trained to sing in the style of jam radio jingles, especially if it could swear...

[LeeCavanagh]

Lol at the jam jingles bit,

I feel it’s going to be everything I hate from local radio , interested how good the voices can sound to an original with personality inracked tho, purely educational.

[Polecat]

Seems to be just another step away from the essence and appeal of great radio. The human connection is what makes it addictive.

Looking at the popularity of some podcasts, where real people talk about real issues or interact with others in an entertaining way, it’s easy to see where radio went wrong.

I look forward to the day when the FM dial is just pirates doing their thing.

[Albert H]

Automated "Davy Doubledecks" - I couldn't think of anything worse!

[tvtv]

Automated "Davy Doubledecks" - I couldn't think of anything worse!

Absolutely, it depends how and when it’s used, like anything. So many stations are automated or contain repeated material.

[radionortheast]

In south korea it may work, they have computered generated k-pop groups, here its a bit alien, it would like letting influences on air, just abit to left field, lot of stations now have presenters who are musicians, ex tv presenters, their feeding into that nostalgia for the 90's, someone from the all stains is someone that you know, the old ILR stations have to change be more active online and be doing things, they can't change to much.

[Albert H]

Automated "Davy Doubledecks" - I couldn't think of anything worse!

Absolutely, it depends how and when it’s used, like anything. So many stations are automated or contain repeated material.

... Of course the "voice tracking" that's widely used by the less-good radio stations is close to the automated DJ rubbish that ChatGPT and the other LLMs are generating. (They're NOT "AI", by the way - they're just crude language analysers).

Way back in the late 80s, when the IQD digital audio recorder chips became cheaply available, I built (as a joke) a DJ-replacer with a load of samples of a well-known BBC Radio DJ. It could deliver about 40 phrases, and with a bit of timing and dexterity, it wasn't possible to hear the difference between the sampler and the real DJ (he was notorious for repeating himself between tracks, and had about 15 "catchphrases" that he used repeatedly!) OOPS!

[Albert H]

OK. We tried out a Text-to-Speech tool on a live programme recently. We'd "trained" a speech cloning "AI" tool with the presenter's voice (actually - mine!). Live on the programme, after a couple of conventional inter-music links, we started typing the links and playing them in between tracks!

Listening back to the ROT, unless you're listening very carefully, there's no real difference. The timing of phrases seemed a little "stilted", but that just sounded like I was distracted whilst talking.

At the end of the 2-hour show, we asked the listeners if they'd noticed anything really different about the show. One listener said that there were fewer "errs" and "ums" (that sometimes pepper my delivery, alas) and he thought that it might have been a recorded and edited show. Everyone else who commented pointed to differences in the music content!

In the event, it was a useful exercise (though our typist at the keyboard had to work hard!). I could see me using it legitimately if I had a bad cold and a long programme to do - it's a way of not "going off sick" - but in the general run of things, I much prefer the imperfections of a real, live delivery from me (or almost any other competent presenter!).

Unfortunately, I can see stations like "Heart" and "Galaxy" and all the other, similar dross using this technology to replace their DJs. Looking at a Heart presenter contract, I notice that Heart "reserve the right to record your output for other, further future use", which I took to mean the execrable "Voice Tracking". With the advent of these LLMs, I can see the forthcoming demise of live presentation in an effort to reduce costs (and thereby increase profits!).

[LeeCavanagh]

There is only one app i like, and have use oreate some idents, not used it to do anything more than that but what i have used is pretty goodut there are toomany bad attempts at using this tech on the app store, i know as i tried all the trials, but one has some VO voicestht sound ntural and you can cgange the style , exageration etc