'AI is the vessel.' A 'long-winded' chat on AI music with Fresno drummer Steve Cox.
Plus a concert-announcements list. It includes Joyce Manor.
Steve Cox has been experimenting with AI in his music making over the last year.
Using the Suno platform, he’s brought to life a trio of projects that may have otherwise been indefinitely backburnered. All have releases streaming now.
“I feel like there are some very common misconceptions in regards to AI music,” says Cox, a Fresno-based drummer who reached out after reading my post on Xania Monet last month.
“The big take away here, and something I truly didn’t realize or fully understand when I started on this journey, is, AI is a tool.”
Here is our conversation. Answers have slight edits for sake for clarity and/or brevity.
OK, walk me through the projects on which you’ve used AI.
Athens to Ashes - City Of The Dead - Original metalcore/numetal band based out of Lemoore.
“The band is currently being rebranded, but, what’s available to listen to currently (more coming out soon, it just hasn’t been published) was all written and arranged by Josh (Davis) and myself back in 2020-21.
We recorded the instruments live in studio.
It was during COVID, and there was still a lot of social distancing going on. So, some of the recordings were done one at a time, or, like in our the case of our vocalist, recorded remotely.
Those songs were never released.
We made the mutual decision to part ways with our vocalist, and as is the case with a lot of bands, finding a new vocalist became a problem. After auditioning several, for months, we decided to go on indefinite hiatus.
Fast forward to 2025, and a friend turned me onto an AI music generator called Suno. After playing around with it for the weekend, I realized what a powerful tool it was, and how it could potentially help solve our dilemma of finding a vocalist.
I uploaded our stems [the recorded audio tracks] into my DAW [digital audio workstation], bounced a WAV file with no vocals, and input it into the Suno software. After a few weeks of training the AI, I bounced a working demo to the guys and asked them what they thought.
They were both super impressed and stoked, but wanted to know who was singing.
Well, technically, it was me, but not really me. The voice you hear on the finished song is me, but with singing from an AI persona that I trained.
The AI was also used to refine the song and make it more polished, adding the 808 sub drops and dj scratches for that numetal vibe.
The stems are redownloaded back into my DAW for the final tweaking and mixing, and mastering before uploading for distribution. So the music, the lyrics, the arrangement, all of the ideology behind the style and everything, is all us.
The AI is in the vocal sample replacement and the production.
The AI also reamped the guitars, and used different samples than the original drums. But I can still show the original raw stems to compare that the playing is essentially the same.”
Echo Of The Unseen - This is my solo project.
“It originated from songs I had written years ago for Athens to Ashes that the band decided were either a little too risque or didn’t really fit the vibe at the time.
Some of these songs are very personal and deal with past relationships, views about my faith, politics, etc. We live in a pretty conservative part of California, and there were concerns some of these lyrics would be offensive to a large part of our fan base.
Some of these songs were written using a synth and a midi controller, as I don’t play guitar (think ala piano roll). The AI allows me to write something on keys, and then sample replace that to sound like any instrument I want.
I’ve also used Guitar Pro to program guitar parts, or even sing guitar riffs into a voice memo on my phone and turn them into songs.
Most of the drum parts are still written and either performed by me, or programmed, depending on the vibe I want.
Vocals are done the same as in Athens to Ashes. All my lyrical content is done by my persona that I built and trained from my own voice and then replaced the same as if re-amping a guitar in post.
This project relies more heavily on AI due to the fact it’s a solo project and I don’t physically play some of these instruments.
Occasionally, I will use the AI to prompt a feel I’m looking for, and use that as a stepping stone or jumping off point. Often, it’s still me prompting the AI with tempos, key/tuning, chord progressions, etc. The song structure is still all my arrangement as well.
So, the AI is used more in the creative process when it comes to the music. But it’s still my words, my emotion, my experiences being told in the story.”
False Equivalence - This is a similar project in terms of its AI contribution.
“It’s a hybrid country/numetal genre-bending band that I hope to actually form with real players. Think Hardy meets Linkin Park. Country roots, with angsty energy and passion of numetal/metalcore, with some trap beats, 808s, scratches and rap thrown into the mix. All while still rocking the fingerpicked acoustic guitar, banjo, fiddle and mandolin.
The project is acting as a test bed to write demos and be able to shop interests. Then, I’ll use those songs as a means to audition real live players. Once that has happened, the next step will be booking live shows for that band, and re-recording everything with actual players and vocals.
Same thing though: my lyrics, my vision, my stories. The AI is the vessel that is being used to accomplish this.”
What was the initial process like? Were there tutorials involved? Was it all just trial and error?
“The original process was a lot of trial and error. Watching Youtube videos, joining AI music community groups on Facebook. I was pretty much anti-AI anything until this project opened my eyes.
It was on me to really sell it to the guys, and make sure what we were doing had the right vibe and was something we could be proud of. One of the first things they both asked me was ‘Is it AI singing?’
After I explained the process of developing and training my own AI persona for it, it began to click for us all that maybe we just stay a studio project, continue writing, sharing our stuff, and letting the process take shape with this new direction.
Less stress and hassle of finding a new singer, more creative control for us. Based on some health issues and issues in our personal lives that all seemed to be the best path forward at this time.
So, yeah, lots of trial and error, lots of late night/early morning emails/texts, lots of listening to the same song over and over and over, and fine tuning everything until all three of us were satisfied with it.”
Do you label the stuff you put out as being enhanced by AI? Does it matter to you if people know?
“I don’t label it as AI. But I also don’t hide it. I’m pretty open about it to anyone who cares enough to talk about it. I frequently refer to my songs as ‘AI assisted.’
As for credits, all of our stuff, and all my solo stuff, is all registered with BMI. All of it has been submitted to AI screening/detection software, analyzed, etc. and has passed any and all quality control standards set by all the major streaming platforms without being flagged as AI generated.”
Is there a line in which using AI goes too far?
“Absolutely. In the process of trying to figure all of this out, there are certain things that come to light.
Like anything, there are certain people whose only intentions are to get rich quick. They’re always looking for that easy cash cow that they can manipulate for a quick buck. There are a ton of people out there gaming the system, using other AI-based software to create entire albums at a time, for the sole purpose of uploading as many songs as possible and hoping something hits.
They’re throwing crap at the wall to see what sticks.
These are the people posting 100+ songs a week, most of which they haven’t even listened to. These people dilute the market and make it harder for people making anything good to be heard.
These are the people the streaming services are cracking down on; AI generated prompts, making AI generated lyrics, that then make AI generated songs with zero actual human creative input.
These are the people most real artists are pissed off about. I get that and I don’t blame them at all.
The other reason people are pissed is the lawsuits you’re seeing, where certain AI generative software has used real artists in its large language model in order to train the voices and musical styles it is generating.
A lot of people see this as cloning, copying, etc. and feel it’s a violation of their rights, their copyright, and their intellectual property. Ultimately, that leads to money problems and artists basically not being paid royalties for their input, which they never granted in the first place. It’s piracy in a sense. Napster all over again.
But all through musical history people have scoffed at technology. The microphone, the guitar amplifier/distortion pedal, Electronic drums. The use of sampling, modeling pedals. Presets and plug-ins in your DAW. Ultimately, none of those things work without human creativity and input.
AI is no different.
Download Suno, and then don’t do anything with it. It won’t generate any songs. Because you haven’t told it to.
If the argument is AI generated music is crap, it’s because crappy people are using it. How is that any different than 90% of the rest of the world’s musicians who are out there making ‘real’ music? Most of that is bad too.
Good musicians will continue to find new ways to adapt and use the tools at their disposal, to over come and make new, GOOD music.
Music is an artform, and no matter how good the computer models get, they just don’t posses the same level of creativity that humanity does.”
Mark the calendars: Chris Isaak, Joyce Manor, Powerman 5000, Sponge
It’s been a week for local concert announcements.
New Rock 104.1 booked Joyce Manor for a Dec. 16 concert at Tioga Sequoia’s beer garden. This is part of the station’s Sound House series, which means the band will be doing a stripped-down acoustic set of five or six songs. It also means the show is free and open to all ages. Show starts at 7 p.m., thought fans in certainly be lining up early. Entrance info to come.
Chris Isaak announced a Feb. 25 date at the Warnors Theatre. This is the second (third??) show announced at the venue since its partnership with Ineffable Live and is a sign of the type of concerts that might be follow.
I was reminded that Isaak played the Cadillac Club in the early ‘90s (though I was not … old enough to be there).
A pair of concerts were announced for the Sanctuary in 2026. This is the new venue on Fulton Street in what used to be Full Circle Brewing Company (and before that Zack’s Brewing).
The ‘90s alt rock band Sponge announced an April 12 date via Sucka Punch Productions.
The promoter is bringing in Powerman 5000 the following week (April 24). That show will feature support from Adema, Makes My Blood Dance and Fresno’s own Mindbender (back from Europe).
That’s it for this week. Remember you can also hear me on the Homegrown Show Sundays at 8 p.m. on New Rock 104.1 FM. Tonight, I’m in studio tonight with a bunch of new music, including a single from 123 Death. Follow my other writing at The Fresno Bee. If you have anything you think I need to be looking at or listening to, feel free to let me know: jtehee@gmail.com