Jammin' With AI

 

In this post, I decided to describe my experience creating music with AI, namely with Suno.com. I must say it was both exciting and frustrating at times. There were ups and downs, yet overall, it has been an interesting endeavor.

I also understand that this topic may seem controversial to many. Artists often feel threatened by this new technology. I think, however, that these worries are rather unfounded. I will try to explain why.

Note also that I am not trying to advertise Suno or similar AI tools. I am simply seeing them for what they are — very useful at times, sometimes necessary, and very often just entertainment for casual users who think they can create music.

Let me say right away that for many non‑musical songwriters, Suno and Udio open a unique opportunity to turn poetry into songs. The songs, however, are usually not mixed with professional quality. It also takes real musical experience to write a good prompt with appropriate arrangements. Thus, most AI‑created music is not serious enough, does not go anywhere, and will not challenge real bands any time soon — I can guarantee that.

You have to be an experienced producer to properly finish an AI track; otherwise, it won’t sound like real music. On top of regular mastering, one must more or less remix the whole thing. This is necessary to find balance, create transparency, and achieve a tolerable dynamic range. And that’s not all. You also have to humanize the robotic vocals.

Finally, AI sees music as noise — but not just any noise: the kind of noise humans like. Because of this, AI creates an entire song in one shot without making stems first. Hence, any attempt to seriously use the stems that come with the mix is hopeless. These stems are generated from the full song, created with losses, and it is impossible to reassemble the track from them. Maybe they could be used to generate MIDI files and rework the whole composition — I didn’t try this. But then, why so much hassle? In such a case, you might as well just make your own real music.

And speaking of noise, you will have to clean up a lot of noisy resonances by notching them out — sometimes partially, other times entirely; sometimes static, other times very dynamic.

All of this is tedious work that takes a lot of time and skill. Occasionally, you can get lucky and receive a reasonable mix. Then re‑mastering won’t be so bad. But it is almost impossible to guarantee a good mix, even if you opt for the most conservative, mainstream arrangements.

One clear example of AI seeing music as noise is the always‑problematic low end. It is consistently noisy and requires extra care, using the right compression and dynamic EQ. Dynamic EQ is often necessary because there is no real separation in the mix. To tame vocals, for instance, you may still need to tame the low end — where, under normal circumstances, you wouldn’t expect major issues.

Thinking allegorically, I see real human music as something smoother, where the direction of adjustments is more predictable than in a typical AI mix. With AI, the smallest adjustment can throw the balance off, kill transparency, or even make the sound unmusical. Hence: save your work often. Save again when you enter a better‑sounding regime.

Given that you may end up with a very long mastering chain, you might eventually have to render your work and then continue mastering the rendered track anew. You’ll get a kind of “twice‑distilled” sound and, hopefully, achieve acceptable quality. Still, this is not a one‑size‑fits‑all approach — it is very case by case.

I am writing this not to discourage anyone, but to let musicians know how challenging AI music creation can be, while also giving aspiring AI artists a realistic idea of what to expect on the path to success.

I should also mention that I’ve been receiving a lot of AI music submissions lately and have rejected most of them for one basic reason: lack of quality. I can think of only a handful of artists who consistently deliver great results, and I will mention most of them below.

In other words, the potential is still unclear. The hype is real, but what place AI music will ultimately claim remains to be seen.

Finally, for those who think radically and want everyone to boycott this new technology: the big three labels are already moving in favor of AI. The litigation of Sony, Warner, and Universal against Suno and Udio is roughly 50% settled, and I believe they will eventually find ways to work together. In the end, these labels need hype too — and AI tech is one of the strongest sources of hype today.

 

News

A while ago, I decided to try creating AI music. I was inspired by my collaboration with Stefan Michael Savolyi, who provided five soundtracks for my audiobook Fenrir – The Spirit of Freedom. I’ve got to say: AI was the only realistic way to accomplish it, given the complexity and magnitude of the project, as well as tight deadlines. I finished the book in one week while the muse was strongly present.

Right after that, I got very lucky completing my first AI project, which I called Jammin’ With AI. I say “lucky” because re‑mastering the sound was not that difficult. Things became more challenging when I pushed the limits and tried something much wilder in the next three singles. That experience helped me understand both the limitations of the creative process and what can still be achieved by mastering less‑than‑great mixes.

Recently, I released many AI compositions in various styles and genres. Most of them are featured on my brand‑new playlist Jammin’ With AI:

https://open.spotify.com/playlist/6xCis6HyviA9HfzhDhEET1

Some compositions were too long for a playlist. You can find them here, in the same order:

Deep Sleep Spa Relaxation (Canadian Zen) • Organ‑ic Dinner Jazz • Chillout Lounge Dub:

https://www.youtube.com/watch?v=vgko5ntrlFI

https://www.youtube.com/watch?v=FAzNpKiuFTI&list=OLAK5uy_mpVjXQdLClaNtpnC37EnXA5XChHUtrgTs

https://www.youtube.com/watch?v=n7PujGg4ivU&list=OLAK5uy_mitKzgqAgRdpO0yjVoEPtEfldtBRQykE4

 

Music Picks

Lately, I’ve been receiving many AI music submissions, and I’ve developed a pretty good instinct for spotting AI‑generated music — especially when the artist hasn’t re‑mastered the sound. In such cases, it’s an obvious reject.

At the same time, I’ve come across some truly remarkable artists who put real heart and soul into their work. I’m happy to present their music below.

All of the music mentioned here can be found on my playlist Jammin’ With AI — a playlist dedicated to high‑quality AI music. There is no, and will be no, cheap or poorly edited work on this playlist. I believe it’s important to showcase only well‑produced content with genuine artistic value. On one hand, this shows that AI music can sound excellent; on the other, it demonstrates that achieving that quality still requires work, dedication, and experience.

I’ll start with music by my collaborator Stefan Michael Savolyi. First, you’ll hear two remarkable instrumental compositions, followed by the opening and closing soundtracks for Fenrir, and another instrumental metal track.

AI‑KI.1 Paradisus fictus – Elysium

https://open.spotify.com/track/4APeUGpaOeOjWH0Nlb5P0E?si=c12946b242f94247

 

AI‑KI.2 Chordae caelestes cum clavile – Angeli caelestis

https://open.spotify.com/track/4vlK2bEfpBvOZHkiFPMzW5?si=d870ad25be8d438e

 

ProphecyFenrir – The Spirit of Freedom OST: 

https://open.spotify.com/track/2ILnZgBOzBXf63FcWLohVA?si=f0f18502c82b42ee

 

The SagaFenrir – The Spirit of Freedom OST: 

https://open.spotify.com/track/2BHQRFjJIE26lpMGnU6VJn?si=27a52402f8c0497c

 

The Beginning – instrumental (project Decanon): 

https://open.spotify.com/track/4AvKoTVgC4o5P3CL1HpcrO?si=d6ac0e12e9ff4f73

 

Next, I’d like to highlight another incredible AI artist who not only creates compelling music but also produces spectacular music videos: 821R (@coordi2025ai). See for yourselves:

Second Phase; TimeTravelMusician; and a couple of songs with titles in Japanese: 

https://www.youtube.com/watch?v=pt5d2Xbp7Pk&list=RDpt5d2Xbp7Pk&start_radio=1

https://open.spotify.com/track/0mHcTETpFiDUebbCO9o4sl?si=1f9da121aa054b9f

https://open.spotify.com/track/74eTUduKZa3B7KPq4iowat?si=dad5b6532d8d405a

https://open.spotify.com/track/5qBUoBcgyUcOUZIr85NT6n?si=4d9f417a379e49a7

 

Here is another song that was previously mentioned on my blog. I love the lyrics, and I’m glad AI gave the artist the opportunity to turn them into a very cool song (Amen by Circle):

https://open.spotify.com/track/6yaOPLIj5ZzOHlZ1Siq5fu?si=16313ba89369413f

Not every artist is ready to publicly stand behind their AI work, even when the quality is exceptional. Here are a few more tracks worth your attention:

 

You Were Never Mine by Novi: 

https://open.spotify.com/track/5BzpINdshSDfqPBmMTruMr?si=5d138b60633a46ad

 

You Play The Game by Def Alfie: 

https://open.spotify.com/track/1Bk4pfRhl3AYsFWD5KoUcV?si=1ce0d3fb9e904177

 

Waiting For The Weekend by Imoliver: 

https://open.spotify.com/track/7evUGnBVDVXPWPbpNztVm1?si=5a34532dc5f64455

 

The Wind Still Knows Our Name by The Velvet Sundown: 

https://open.spotify.com/track/0L43pYkHAh9Tz7UuMl7Jf4?si=1d60fadbabcc4ebe

 

Smoke Talk by JahBeard: 

https://open.spotify.com/track/68W9Mhvx3vsFEUHoJED6e5?si=915f99b8004b4a75

 

Final Thoughts:

There is clearly much more to AI music and production than could ever fit into a single blog post. I will keep exploring this topic and write about it again in the near future. I didn’t want to overwhelm you with information, nor did I intend to encourage or discourage anyone. My goal was simply to share thoughts on a new artistic direction that is actively taking shape.

For me personally, as a prolific songwriter and producer, AI has already opened up unique opportunities that I wouldn’t have had otherwise. To those who criticize this approach, I can point to countless examples throughout history where new technology enabled artistic creation and inventions that would not have been possible before — things we all benefit from today.

In music, even the use of DAWs allows most of us to create from the comfort of our homes without going financially broke or depending on big studios or labels. AI tools like Suno or Udio are simply another technology — nothing more, nothing less. They are not an answer to every question, but they can be useful at times.

As always, I’d be glad to hear your thoughts on this subject. If you’re an experienced AI artist, I’d especially appreciate a comment on this post. Otherwise, feel free to email me or reach out on Instagram. Don’t hesitate — get in touch, and stay in touch.

3 comments