Challenge,  Misc,  publishing

AI Art and Audio…

Topic of a Lot of Discussions Among Writers…

Lots and lots of early adaptors in the indie writing world are trying AI.

For audio, the technology is jumping ahead at full speed and the prices are slowly coming down. For nonfiction, I like the idea and think the quality and technology is almost there now. For fiction, I personally am still on hold until the AI can use my voice and other issues are fixed. But I do know a lot of writers who are trying books with audio. Still too expensive in my opinion, but it does save time. Also limits markets at the moment.

The AI art for book covers is also booming and wow have I seen some beautiful art. Wow, just wow. The problem here is the copyright issues. Some basic issues have been cleared in the last few months, but there are still major copyright issues on the art used to train the AI. I doubt the issues will come through the programs to the users. But again, for me, I am in a wait and see mode.

But got a hunch I will be playing with a few programs in the not too distant future.

Fun to watch, fun to play if you are an early adaptor. And I don’t mind reports on either. Especially cost and time on the AI audio and any copyright cases on the AI art.


    • dwsmith

      This is a fantastic article. I would suggest everyone read it to see the full scope of the problem and the direction things are heading.

      Thanks, F.I.

      • Cheryl

        Thx, F.I., for that resource.

        I am reconciled to the fact that I’m relegated to Cast Aside In The Dust status with the stampede to All The Wonders AI Can Bring, but I am generally not a fan. I recognize I’m somewhat hypocritical, because I use Affinity Photo (Photoshop replacement) for compositions, and I don’t feel (too) conflicted with Dean’s thought of using AI voices to replace skilled voiceover actors (thereby taking food from their mouths). Perhaps it’s this swift slide along a spectrum from human art and craftsmanship to a few minutes of digital input. I’ve no doubt it’s colored by my own respect for “analog” artists (my husband and I met in art school eons ago). So not like I’m going to put my hand up and stop a runaway train, but just in case anyone else has similar thoughts, I thought I’d give voice to them.

        • dwsmith

          The AIs have to be fed and programmed and the artists feeding the AIs need to be paid. The question is going to come down in courts if the artists, like music, get paid by the play or not. Going to be interesting to watch it all shake out. But most of it will be in courts.

          • Desikan

            Hi Dean,

            Coming from the tech side, these models once trained fully, will then need to be retrained only periodically with further incremental data. I would expect this to end up with a need for less artists overall in the long run.

            However, I feel human creativity is endless and always will find something more interesting to do with any new technology.

  • Mark Kuhn

    I started playing around with AI art generators a couple weeks ago. I was gobsmacked by the results I was getting by typing in a few descriptive words. I also use it to generate story ideas. A character in a setting with a problem starts it off. I entered something like (I forget the exact terms I used) a man in black fatigues emerges through a doorway of light. Then I chose the fantasy style and hit create. I found I needed to enter terms and words that are similar like, emerges through a door of light, enters an opening of light. The more words the better. It took just a few minutes and I damn near fell off the chair with the image I got. It came back with a black figure standing in front of two light sources in a thick forest of bright colors. This helped immensely with fleshing out the character’s problem. He was stuck between worlds.
    I could be wrong here, but AI images might eliminate duplicate covers from different authors. If you do a search for space opera in the Kindle store, you will find the same space ship on book covers from multiple authors. The pool of images to choose from is large, but limited.
    Then I tried another one. A Red Ninja holding a sword in a dark alley. I used sword and katana. Well I did get a Red Ninja carrying a katana, and instead of a dark alley he was standing in a dimly lit hallway. Cool. The only problem was he was wearing pirate boots instead of tabi shoes and his sleeves were loose and puffy. Got a good laugh out of that one since it reminded me of the Seinfeld puffy shirt episode.
    But the first image is one that I would not hesitate to use on a cover. Stunning.

    • dwsmith

      The Wired article is only about one of the new programs. There are so many, I have lost track to be honest. I doubt highly that any kind of standardization has come about with any terms of service. Just watching is a good plan at the moment.

  • Brad D. Sibbersen

    Nightcafe claims to source public domain images for its basic text-to-image process, and the style evident in the results seems to support this. (They often look like Victorian era – or earlier – art.) It spits out some nonsense, but I’ve gotten at least two images that I’ve been inspired to write stories around. Worth playing with just for that, even if you don’t use the images themselves!

    • Brad D. Sibbersen

      Just a quick addendum: Ignore my comment above and read the terms on the site. Text-to-image creation doesn’t work quite the way I thought it did. I have officially signed up for an account and I can’t recommend it highly enough, even if only to inspire your writing.

  • A Johnson

    It’s funny you mention this, because a photography/art site I visit just had a survay today as to whether to allow AI art on the site. The copywrite issues are definaitely a concern.
    I’ve played with a few AI art programs, but it’s not there yet IMO. You can get some interesting stuff, but they can’t do faces well. I am interested to see where Ai for audiobooks goes.

  • Xander Koolen

    If you want to see some the amazing possibilities of AI art, check out this video from Corridor Crew:

    They manage to input their own pictures (of themselves) into one of the AI art generators and use it to tell a story in a graphic novel style.

    They also touch upon the wider implications of this technology for the art industry.

    • Amy

      Thanks for posting the link to that video, Xander – the images were incredible and the guys presenting it are very engaging and funny. I’d highly recommend it for a watch!

  • T Thorn Coyle

    AI Audio:

    Meant to answer this yesterday, but got a bit distracted by my new Kickstarter launch!

    Several months ago, I put a few of my novels up using Google Play AI voices. IMO, those voices are pretty dang good! It took me all of 15 minutes each, and that included fixing a couple of words I knew the AI would have trouble with (this is trivially easy to do in Google play), making a square cover that includes AI on the sidebar, and uploading.

    I even just used one of these AI audio books as a backer stretch reward for a previous KS campaign. I’ll eventually do more, and put them up for sale on my website. For now, they’re just on Google Play.

    For me, since I’m not going to pay for audio any time soon, the Google Play system is the way to go. And right now? It’s free to use! That will change eventually, so I encourage anyone just wanting to play around with AI audio to do so now.

    Is it as good as live narration? No. But people don’t expect it to be, either.