Radio Drops #26: Thinking About Adding A.I. Music

If you are a radio music director, it is time to begin considering the works of some master creators of AI music. The video runs 6:57.

Watch the video on YouTube: Radio Drops #26: Thinking About Adding AI Music

I’d Never Heard of Her, and She Was Great

A couple of months ago I was in my office, browsing YouTube. One of the suggestions in the side panel was an unreleased blues demo from 1953, with a portrait of the singer: a very large, golden-skinned woman holding a guitar.

It made me think of Big Mama Thornton, one of the blues belters of the late ’40s and early 1950s. She was the first to record “Hound Dog.” She recorded it in ’52, and Elvis released it in ’56, his breakout year. It was the number one record of 1956.

I know a lot of the blues artists who influenced everybody from Elvis to the Stones, and I’d never heard of this singer. So I clicked, and the song is freaking great. It blew me away. It struck me as the kind of record that would have been a huge hit in the early ’50s, or even in the James Brown ’60s.

Then I listened for a minute and thought: this is too clean. It doesn’t even sound like a cleaned-up copy of an old record. And it’s not. She’s the creation of somebody who knows how to tell AI: here’s the kind of voice I want, here’s the kind of beat, do I use horns, whatever, whatever.

Eleven Songs on One Chart

Two or three weeks ago, an AI artist by the name of Eddie Dalton was number one on the iTunes Top 100, with a total of 11 songs in the Top 100 at that time. An AI artist.

Radio guys who use the charts to decide what to add will say, “Yeah, but it’s just the iTunes chart.” The iTunes chart can be manipulated, or so I hear. I don’t know whether it can, but so what? People have always found ways to goose a record a little, to get it a three-point jump instead of a one-point jump. But 11 songs on a chart at one time? I do not think you can manipulate it to that degree.

The charts are not going to tell us what to do. They just give us information to make better decisions ourselves.

Here’s a piece of information. Eddie Dalton’s top song has just short of 3 million views on YouTube in the last month. And if you open Spotify right now and type just three letters in the search bar, Eddie Dalton comes up number one, ahead of Eddie Vedder, Eddie Rabbitt and Eddie Money. Take those data points together with 11 songs on the iTunes chart, and that is an indication there’s some value here.

Ella Langley, Only It’s Not

Ella Langley is the hottest thing in country music right now. About two weeks ago, around April the 10th or 12th, YouTube fed me a new Ella Langley song that’s not Ella Langley. Two, in fact, and to my ears they could be hits.

If I were programming a country station today, I would want to play these records, because I’m in the business of playing hit songs by hit artists. Maybe they wouldn’t become hits. But the data points and my professional ears tell me they’d be worth trying on the air to see if the listeners respond. And if they did, all the better for my station.

From what the good old AI itself has told me, it’s not presently illegal to play an AI-created song on your radio station. And if the argument is that it’s somehow polluting the Ella Langley pool, I don’t know that it is. Does it take money out of her pocket? Well, no.

One of those songs strikes me as a perfect Ella Langley song. If it were to go viral and become really famous, I could imagine Ella singing it in concert somewhere down the line.

Radio Isn’t the Discovery Medium Anymore

From the start of commercial broadcasting in the 1920s to the end of the century, radio was the music discovery medium. That began to change a little when Napster was released in 1999.

You know the lay of the land now. People don’t come to radio looking to discover new songs, like I did, like previous generations did. They discover new music the way I do now: YouTube, Spotify. But it’s still a big treat to hear a new song that really grabs you, no matter where you first hear it.

As a programmer and consultant, I always wanted to position my station as the one that plays the best new songs first. The problem is that people will tune out a station because of a song they don’t like quicker than for any other reason.

Adding new songs, even on a current hits-based station, has always been a chancy thing. You don’t want to play songs people don’t want to hear, and you don’t know what will be a hit before it airs. With all the data points available to us today, it ought to be a lot easier to pick the thoroughbreds out of the herd first.

Play It, Label It, Follow Their Lead

To pick new music for your audience, study what the most active members of your audience demographic are doing. What music are they consuming right now?

If I were a rhythm and blues station today, I would be playing Eddie Dalton. The people who have discovered his top song, the one with the most streams, love it. I’d like the people who haven’t discovered it yet to hear it on my station first. And every time I played him, I would identify him in some way as AI Eddie Dalton.

If I were the country station that wanted to play one of those songs, I would call her AI Ella, clearly identifying for my listeners what I was playing. And then I would follow their lead.