We are inundated daily with news about AI and I know how overwhelming it can be. Most articles make it sound as if we seconds away from Hal flying us into a monolith. I feel my role is to tell you about the fun stuff — the AI that is offering us new ways to be creative and express ourselves. Many of us are worried about AI taking over our professions, but what if it actually allowed you to expand who you are as an artist?
I am an (extremely amateur) musician. I love to play guitar, write songs,and sing, although my singing leaves much to be desired. You already saw my experiments with the video translation software HeyGen which created a video of me speaking perfect French and Italian. Last week I learned about AI that allows you to clone your voice and make you fabulous at singing! (or so the apps promise).
I started with the program Kits.Ai, first watching a how to video on YouTube. The AI can separate the vocals from music and then substitutes any voice you want. The app has dozens of voices to choose from. Here are four:
I was more interested in the feature that allows you to upload your own voice and plug it into any song. How cool to hear myself sing songs that are way out of my range or ability!
First I uploaded a sample of my voice. The minimum is 10 minutes with a recommended sample of 20. I only did eleven minutes and sang into the Memo app on my iPhone, so far from professional quality. But I figured it would be enough to get an idea of how the app worked.
It took over a hour for the AI to upload and replicate my voice (come on AI, get with the program). I then uploaded a couple songs from Music.
Here I am singing “Boys Don’t Cry” by the Cure.
I also tried “Here We Come Again” by Dolly Parton. I knew this song was too high for me, but I figured the AI would just raise my voice - I mean, it is a fake voice. Why not make it better?
The beginning sounds pretty good and I was feeling really inspired to try this song in real life, but then “I” hit that first high notes and “my” voice cracks, which totally made me laugh. Hey AI, you don’t have to rub it in! Maybe it would have been better if I’d uploaded a slightly longer sample that included me trying to hit the highest note possible?
Another things I was very interested in was putting a truly amazing voice into a song I’d written. Here is my original song called “Don’t Call Me Boomer” recorded on my iPhone during band rehearsal (The vocals are a bit drowned out by the instruments.)
I chose an AI voice called Male Pop Punk to replace mine. Here is the result:
So . . . uh, not perfect. Total mush mouth with nonsensical lyrics. It seems the AI is not ready for songs that don’t already exist.
Before my musician friends come at me with torches, let me say that I fully understand the dark implications here that one day professional musicians and vocalist might be obsolete. You all know that I fear the same thing for writers.
But it turns out that in the same way that I don’t fear AI written novels just yet (due to my experimentation with novel writing software such as Sudowrite), you shouldn’t worry that AI is going to sing you out of a job anytime soon.
I recommend that you don’t think of AI on the universal level — what it means for humanity. That is too much for our tiny brains. Think about what it can do for you as an individual. What sounds FUN to you? Is it being about to draw? To sing? To speak Chinese? To make a short film?
My mother is a fabulous artist but when she was a kid she had no way of expressing herself. She tells a story about how one day she got a gift of a box of colored pencils and it was a revelation. It had never occurred to her to just ask her parents for art supplies. Her desire to create was too abstract. Maybe we are all currently staring at a blank page and AI is the box of colored pencils - not the originator of the idea but the ability to take it to new heights.
Anyway, I hope so.
Nailed It
The first time I uploaded my own voice into “Boy’s Don’t Cry” I accidentally told it to remove the original voice AND the music. The result was surreal:
The opening sounds like I am heavy breathing into an old phone while beatboxing. The AI also tried to give me Robert Smith’s British accent which is hilarious. I sound like I have a very stuffy nose. PLUS the lyrics are all wrong (“I tried to nap about it, cover it all up with buys . . . Boys don’t why . . . Boys don’t nye”). It’s like a person singing along with the radio when they have no clue what the lyrics are. (When I was in college I met a girl who had always thought the GoGo’s song “Our Lips Are Sealed” was actually about “Alex the Seal.”) I look forward to more AI hallucinations of my favorite songs.
Very interesting. Looking forward to more experiments. Don’t Call Me a Boomer rocks!
This is amazing... the crack in here you come again was brilliant. I almost did a spit take even knowing it was coming. And that creepy weird beat box song is brilliant...