I posted on this thread back in January. I have since spent a lot of time with Suno Ai. On Thursday May 29th my Album "My Friend Parky" will be available on 100+ streaming platforms (if you search your favorite site with my name - David Rabinow- you should be able to find it. It is about 40-50% Ai. I'm very proud of the work.
This is going to be long, Bear with me.

I have Parkinson's Disease and for the last year and a half I've been writing songs about Parkinson's. And the abum is definitely targeted at people with experiences with PD. As my diseased has progressed my skills have diminished fairly sharply. Since I'm a guitarist I'm am most concerned with how I play on a recording, but I just can't play the way I used to.

I stumbled onto Ai music while I was starting to record some of the songs. I created a voice clone on Ace Studio from my voice and then altered it to be something that doesn't make me cringe. My plan was to record my singing into Ace Studio and then edit in a very midi like format. Unfortunately Ace Studio seems to have a lot of difficulty with English and the results were awful.

Then I found Udio and Suno. They're pretty similar, but Suno allowed me to use my voice clone. My original plan was to create songs in Suno and then splitting the tracks in one of the many Ai stem splitters on the market and then recreate them in my home studio. But because of my limitations the results weren't very good.

I started using more and more of the Suno stems. In the end most of the drums are from Suno. Drums have always been a major obstacle for me. I've never been able to program or write drums so I have relied on BIAB for drums most of the time. EZ Drummer and Logic Session drummer have never worked for me. Suno's drums are MUCH better than BIAB with one huge exception. The sound quality of the cymbals is awful. But the drums are much better at fills and accents and breaks and al the stuff that you want your drummer to do.

Suno, and all the other Ai music is at an infant stage - actually probably more like a three year old - because it doesn't really do what you want it to do. Typically, I take my lyrics and input them into suno and uploaded an audio clip of me singing the first 30 seconds of the song. I add a bunch of prompts (which take a while to understand) fo Genre, BPM, Time signature, and descriptors for mood. In the lyrics I put in prompts for verse, bridge, chorus, etc. But also performance prompts, i.e., soft voice, harmony, rising intensity, pause, and such. Then I hit create and in about 30 seconds I have two versions of my song. In a few very rare cases, I've gotten what I wanted the first time. Suno doesn't take direction very well. And sometimes I have to run thirty or forty versions before I get what I want.

But here's the cool part. Sometimes Suno's ideas of how my song should sound is better than my own ideas and opened up new directions. After a couple of weeks I started to think of Suno as being like the guys in my band from a previous lifetime. And then eventually I started to think of Suno as a collaborator.

In the end I was able to cull 11 songs and make the album. A critical listen will expose all the flaws in Ai music. But you won't hear them in a casual listen. When the prompts and lyrics overcome the Suno engine it spits out little artificacts. I have done my best to erase them with Melodyne, but the artifacts are often imbedded in the actual note. I'm pretty sure Izotope Spectral Editor could do the job but that's out of my price range. But for me a lot of the artifacts are there.

As to the ethics of using an engine that is trained on using samples from artists without attribution or renumeration, I have to say that as a blues guitarist I was very good at regurgitating what I learned listening to B. B. King, T-Bone Walker, Buddy guy, Mike Bloomfield and dozens of others. In some circumstances that only appears as influence while at other times I WANT to sound like B. B. King. And, Suno claims to have hired and or payed musicians to train their engine.

As far as Ai being some kind of cheating, I have to disagree. We all use tools to make music. But what if, like me, you don't have the skills to play music, but you have something to say and you want to say it musically. In the early days there will be a lot of awful Ai music and some of it will be on the top of the charts. Eventually I believe that the true talent will rise to the top.

It seems possible to me that the next George Gershwyn or John Lennon could be someone who is disabled in some way but now has the ability to express their genius talent they were born with. I don't love B B King because of his technical skills (although if you look on youtube yo can find some examples of him really shredding), I love him for they way he expresses himself.

And finally, a lot of us are using Ai tools in our music workflow. I use many Izotope products, almost all of them use Ai within their features. I've been using EZMix for years and the latest version uses Ai expensively. Ai is here and it's not going away. You can ignore it for now but eventually you'll have to embrace it. I hope that PG Music is looking at how they can integrate Ai in BIAB. There may be a perfect song creation program that's half BIAB anf half Suno. Who Knows?

Thanks for reading