The story behind the story

This week’s newsletter is brought to you by doing something completely different, but I’ll tell that part of the story at the end, so the payoff is fireworks-sized. No links in this, other than a link to listen to a podcast episode and the totally true story of the editorial decisions and labor that went into it.

Those of you who have been following along at home are already aware that I’ve been working as the editorial director on a new podcast called Have You Heard This One? Melissa Locker invited me to come aboard the project, and we’ve spent the better part of 2023 working on the launch, learning a lot of lessons, and making some fun episodes with wonderful writers and a stellar production team of a fully immersive show that tells stories lost to the sands of time about music history — and that frequently means covering underrepresented people and cultures.

Last Wednesday, we published an episode titled “Who the Heck Is Ursula Bogner?” If you don’t know who she is and don’t want the story spoiled for you before you get a chance to listen, stop reading now and come back.

SPOILER ALERT FROM HERE ON

The salient details on Bogner are that she’s a “found” early electronic music pioneer who was making songs from the ‘60s to the ‘80s but never released her music. Through a chance encounter, this German electronic producer named Jan Jelinek met her son, convinced him to release her music in the 2010s, and created a label to facilitate doing so. There’s a lot more nuance in the podcast about her history — but the catch is this might all be a hoax, and Ursula may not exist. If she does, no one has been able to prove it.

Our first hurdle came from needing to imagine Jelinek’s voice. He did not want to do an audio or video interview but did agree to answer some questions over email. Obviously, that’s not ideal for a podcast, but the story would be incomplete without him — and speaking to Bogner was impossible because she’s said to be deceased, and her son has never been located (again, he may not be real). Jelinek is effectively the only source for Bogner’s story. So, we decided to use an AI voice.

The author and host of this episode, Rose Bacci, is a woman with a feminine voice, and Jelniek identifies as a man, so we went for a masculine AI voice. There were numerous discussions from there — we opted not to use a German accent, and we toyed with various iterations of the voice to make it realistic enough but not completely realistic. We wrote a script where Rose introduced it by explaining it was an AI voice and why we were using it in lieu of a recorded interview.

After listening to a version of the episode with those two voices, we all felt that we wanted to hear more of Ursula’s music. The script had already been recorded, so adding more text wasn’t an easy option. I decided to inject quotes from reviews of Bogner’s work on some music writers at websites that are considered serious and frequently referenced — two from Pitchfork and one from All Music Guide — that described various songs on her two albums, after which we played snippets of the songs. There were some layers to the quotes themselves, wherein all-male music writers had these flowery things to say about the songs, sometimes orienting them in time and space by comparing her work to other artists. Laid next to the actual music, however, which is incredibly simplistic, there’s almost a feeling, for me at any rate, of buying into the hoax and imbuing this artist with legitimacy.

We decided to layer another AI voice into the script to read those quotes. We, again, debated about whether it would be masculine or feminine and searched for one that was even less realistic and clearly computerized — like an ‘80s movie version of what a talking computer would sound like, almost. When we found the right AI voice, we opted for a feminine one. For me, the layers of having a female AI that was obviously a computer reading the words written by men on websites describing the music of a woman that was made on a computer who might simply be the alter ego of a man was hilariously meta. We don’t spell that out in the episode, but I seriously hope that some people got it, even if only subconsciously.

On top of that, I wrote a script for the AI to read when it introduced itself right after our host introduced herself, which made it clear we were making a statement by using it. Rose says: “I'm Rose Bacci, and you’re listening to Have You Heard This One? A show about the stories in the back pages and hidden corners of music history.”

And the AI then says: “And I’m an AI voice. I know you’re probably already skeptical about me, what with all of the plagiarism, lies, and deep fakes. But don’t worry, the creators of this podcast brought me in to read real quotes written by real people. You can believe me. Everything I say will come straight from actual living, breathing music writers. By the way, those are all overrated traits for a writer to have, if you ask me. But you didn’t. I’ll only report the facts, or opinions, really, of so-called human writers.”

The aim was to acknowledge the very real and justified angst writers have about being replaced by AI, the discourse (which was heavier at the time) about AI creating flawed and falsified results when asked to do research or write something, and to be funny — and succinct.

The episode turned out to be our biggest creative swing, in my opinion, and I hope we get the chance to do something like it again. And if I’m being completely transparent with you, I had all of this written down already because men on the internet decided to man at me last week after that episode dropped. Not one but two of them reached out with suggestions about how this episode could have more responsibly used AI or questions about why the episode was formatted to incorporate AI and not the humans who wrote some of the material it quotes — and it quickly came to light that neither of them had actually listened to it. The incredible and very particular sense of defeat that made me feel was intense.

They also raised questions about the ethics of using AI, pointing to strikes that have been happening this year and general conversation about it among unionized newsrooms. I am a member of the WGA-East, which I assumed neither of them knew because they are not. I did consider the use of AI, primarily from the point of view of voice actors, which is a constant conversation in our union’s Discord channel. We discussed having a well-known actor read the Jelinek part originally, but it didn’t work out. Once we decided to go with AI there, adding another AI in for the quotes became part of the sound design, the very concept of the episode and the nature of what it commented on.

So, that’s the inside story of a particularly overwrought week on the internet — er, and an episode of a podcast.