Beep Boop Beep: Error Report – Or “Spotify, Findaway Voices, AI Audiobooks, and the Problem of it All”

Firstly, what a wild ride it has been from the Hugos cocking up yet again because gotta love bullsh#ttery. Nice to know I’ll super never be able to win a Hugo, either because they hand out awards to writers of color like Scrooge hands out charity or if I even get on the ballot (I strongly doubt that), whatever secret dossier they’d make on me would be tall enough to double as a barstool. Sad Puppies and now this? Hugos really doesn’t like writers of color at all, apparently. Even the Nebulas (which is by SFWA) has tried to do better, to give credit where credit is due. At least to the point of not throwing ppl off the ballot and pitching a Standard “Whiteness Means Authority & I Am Authority” White Guy hissy fit when people ask about why the ballots are wonky and if there was some backroom dealing at play – even to other White people, which is a bit on the humorous side, I suppose. It does not help the fact at all that one of the people who was part of making secret dossiers and keeping writers of color*, particularly Chinese diaspora writers, out during the 2023 Hugo was allowed to be on the board again for the 2024 Hugos until the huge exposé hit the front page of File 770. She only stepped down – not was thrown out – once this info had hit the stands. In other words, the Hugo committee thought that she was decent enough, despite all the bs milling about her and her crew that had already happened, to bring on again as an administrator for another year.

At least there’s the Ignyte Award.

That’s all a different post for a different time.

This post is about good ol’ AI issues in the arts. As reported by Writer’s Beware** and I can confirm with my own emails I’ve gotten from Findaway/Spotify, they’re getting way too comfy with AI voices and rights grabbing from authors and narrators alike.

Now, there’s no getting around it, narrators are expensive. If you’re recording a 3-5 hour long audiobook using a professional narrator/voice actor, they can cost a stack easily. Not everyone has a grand or more laying around, especially not indie writers. I rely on grants and royalties to pay for mine. I can see how and why an AI voice sounds like a really good option for those who are lean of pocket but far of vision.

It also doesn’t help that audiobook dealers and makers, such as Google Books, Draft2Digital and Apple are pretty much urging authors to switch to AI voice. Only a couple days after the Spotify debacle, did Google Books come out with this email:

Either Google is trying to jump into the fight after seeing Findaway/Spotify or this is some unfortunate timing (I vote the former).

AI voices, for them, looks like a champion choice because it’s their products and it produces the one thing they like, besides money: data. It is so much easier to control content and who gets seen and who doesn’t get seen when your proprietary robots get used. Given that tech companies care less about ethics than the average Hugo Awards voting board, this is a bad thing ultimately for writers and narrators. And listeners because if there’s one thing I’m noticing, AI isn’t as widely received with open arms by everyone.

AI makes data procurement easy, because it’s all ones and zeroes, at the end of the day. If Apple wants to ban content about China or Black Lives Matter, an AI, especially if it’s an AI narrator they created, is going to do exactly that. Google already got in trouble in the past with turning “I’m sad for Hong Kong” in English into “I’m stoked for Hong Kong” in Chinese – without telling the user. This means ethics bending is well within their realm of “Things they are ok with doing”. An AI can’t think, an AI doesn’t have morals. They’re just super impressive YakBaks that will do anything the code tells them to do. And they can censor themselves perfectly if the code tells them to do exactly that. Yes, even without the author or listener knowing, unless someone is reading the book alongside the audio, word for word.

Can’t bring up queer people because you live in Florida or Texas or Uganda or China or Russia? Apple and Google and other companies are fine with turning “girlfriend” or “boyfriend” into “best friend” – even while they pretend they’re pro-queer during June because corporate bullsh#ttery. Or just shadowban the book since it has the “sensitive words” in their digital stores. The AI already knows what it’s going to say so it isn’t hard to check it. That’s their tech, it’s designed to be easily data mined when needed.

And if they can copy human narrators because the company snuck in some mess into the ToS that allows for voice duping, it makes it easy to censor without someone noticing a censor occurred. It’s ok that the narrator said it, the voice double will make sure you don’t hear it and get something different instead that sounds just as fluid. That’s remarkably concerning on a series of levels.

One level is that human narrators deserve to have work, plain and simple. When you’re hearing an audiobook, you’re hearing an audio performance. One that (hopefully) has heart and spirit in the work. At least that’s how I prefer the recordings of my audiobooks to be. You’re paying for a performance, not a cold recite from Alpha-5.

Also, my narrators are hired based on cultural reasons: some words are pronounced or said differently based on context, person and/or situation. Or are just in different languages. Trust me, I have tried AI voices just to see what Google was yammering on about when they first tried to shove it onto me, they also made sure to code in their racism. The words that the AI voice stumbled on were just insulting and agitating and the inflections were plain wrong. And I have to go back and fine-tooth comb the fact the AI clearly had a very White dev team. It’s easier hiring Black, Asian and Latin narrators that can just do exactly what I wrote in the book, no need to “Jan. 6” it up. Like I said, it’s not like I’m ever getting a Hugo.

Plain and simple, the AI sucked because of how much micro-nattering I had to do. It sucks at emotional expression, it sucks at cultural intonation, it just sucks. And if there’s one thing I am diligent about, I don’t like things that cause me more work. If a narrator can’t get an inflection down or emotes incorrectly, I can at least use plain words and me doing a sample take to show how I want things done. And I want my narrators to be within the emotive neighborhood of my take, not always a sharp copy. Can’t do that with an AI, that’s a skill that requires a human touch. An AI doesn’t know if I want someone to wind up into agitated anger because their at their wits end but don’t want to cry so it shows as fury instead, or for someone to give a backhanded compliment as if it were with genuine care. I can work the AI eventually to give me that but it would be too much work that I 100000000000000000% don’t feel like spending because why give myself more work just to hate it all in the end? But a human narrator can deliver that effortlessly, especially when you instruct them plainly to or just show them yourself, in a snap.

Yes, narrators can be super pricy but also some narrators will work with you in the payment department, especially up and coming ones. You don’t always have to sink a rent payment to get a decent audio version of your book. And I have heard AI narration, it really isn’t too awesome. It can sound booming as if told by Morgan Freeman but I rather hear from the original, award winning actor Morgan Freeman. AI voices try to sound varied but in a very constant way. Like they’re forever interested by what is being said and stuck on “forever interested” or “forever dramatic”, if that makes sense. People don’t actually talk like that.

I super see the ease of why indie authors and others would want to use AI voices (and you know trad pub would love to use AI voices if it means cutting someone else out of a getting a paycheck and isn’t a CEO) but it’s a lot more than someone can bargain for, that’s for sure. It simply isn’t worth it.

* They also got White writers mixed up in this Hugo mess but, people, it’s the Hugos, Paul Weimer and Neil Gaiman will get other shots at the Hugo award as long as they don’t suddenly lose their ability to be literate. It’s a bigger uphill climb for writers of color. It took the Hugos 63 years since their inception in 1953 to give an award to any writer of color. That happened in 2016. To N.K. Jemison. If you’re White, you’ll get another chance, just keep writing. If you’re not, that can easily be your one and only chance

**I may have my criticisms of SFWA that are a tower tall because of their past behavior – but Victoria Strauss is a total gem. A walking credit to SFWA, I trust her like Walter Cronkite. Trust is earned, not given – esp. when it comes to broken trust – and this is how you earn it

Leave a comment

Create a website or blog at WordPress.com

Up ↑