AI disqualifies anyone as a musician! It's like playback.

[VOODOO U]

So then what the f**k are we doing here? We might as well just close up this forum and go do something else.

That's like saying we should close up Bandcamp because it doesn't have a forum. KVR would do.just fine without the Music Cafe.

Me? I happen to be damn proud of my songs

Great! As it should be.

and I don't need YOU or anybody else to tell me how great they are.

How do you like them apples?

I would never tell you your songs are great. I'm one of the honest ones here.

[ghettosynth]

How do you like them apples?

[eassae]

Do one about Praga Khan needing his nappy changed.

[wagtunes]

How do you like them apples?

I have to ask. Was this AI generated? If so, what service did you use and just how did you go about the process? Because I gotta tell you, this is f**king cool.

[wagtunes]

How do you like them apples?

So here's what I'm going to do. I'm going to take the title and do my own version. I'll try to keep the subject matter the same but will write the lyrics myself and compose all the music myself. Dollars to donuts, your version is going to be more cool if nothing else.

Thanks for the inspiration.

[Seafire Mk2]

There's that potty mouth again...

Indeed!



Or, if it's time for a little break, enjoy the potty break mix!

That is actually awesome. Love a bit of acid techno

Edit: played it to the Mrs, she lol'd

[ghettosynth]

I don't hold the fact that I don't like you as a person against you when listening to your music.

You already did by saying that

". But "It's crap" is not constructive criticism. It's just being a lazy SOB.

An opinion doesn't have to be constructive, it just history be honest.

Well, I tried. So here's some honesty. You're an ass.

Note, the title of this fell out of Suno's new moderation policy where the word "ass" is too strong. I suspect that we are coming to an end of the "I glued my balls to my butthole again" era. I chose the moment to cancel my subscription which was set to renew in November. Note, I had intended to cancel at the end of the year anyway. In any case, it was fun to play with this year.

Seriously, I can't even tell you how much joy you've brought into my life with your videos man. You got me dying laughing over her. You are the forum futurist! Thank you for that! lol

[Zeisner]

Still sounds fairly realistic to me:

https://www.flowmusic.app/song/081c2f3e ... 959c02007c

The background is not that bad. But the lead is. It sounds very synthetic, even worse than mediocre sample libraries.

[ghettosynth]

Haven’t checked up on it lately… apparently Google’s Producer AI is now called Google Flow. Still sounds fairly realistic to me:

https://www.flowmusic.app/song/081c2f3e ... 959c02007c

So flow.app existed before google, I had an account, it wasn't very good back then. I currently pay for gemini because it's a remarkably capable bundle, hence, I probably won't get rid of it for a long time, if ever, we'll see. However, my account now has a boatload of tokens because of my subscription. So, I will explore as well and compare against what Suno can do.

Both apps are now allowing you better control over track length. I think though, that the flow workflow is superior, if, it's capable of getting you to where you need to go. What you shared there sounds better than what it was able to do at the end of last year or so. I think the google integration helped quite a bit.

AFAIK, both models allow you to extract stems. I expect that to be the dominant feature for musicians wanting to use it to build out song structure and then replace certain parts.

[D-Fusion]

We all need a bit of Irony in our life
https://soundcloud.com/d-fusion-6520430 ... al_sharing

https://soundcloud.com/d-fusion-6520430 ... al_sharing

[ghettosynth]

Ok, so, I created a "version" of the song above in google's flow app. Obviously this is not exactly the same. It was primarily a comparison of process and sound. Flow still has annoying limitations. The model tries very hard to force a three minute pop song length, which kept bumping into the repetition that I used intentionally in the previous version. There are ways around it, but, they are awkward. You can download stems, but, for me, there was an error in downloading the bass. They are broken into just four files whereas suno breaks them into eight. I called it a day and generated the Summer Radio Edit. There are key differences in instrument quality between the two. I'm not going to do this normally, it's not really my thing. In this case wags asked about process and so some of you may find this comparison instructive. My money, long term, is on flow being the better tool. For me, it is essentially free as I pay for gemini pro for other reasons. YMMV and all that.

The key difference is that with flow you are interacting with a language model that will make changes to the style prompts and guide you how to use the tool. One key aspect that helps to make changes while keeping a similar output is that you can control the seed. However, note that if you make changes to the lyrics that it regenerates the song and it finds a different path through the latent space, so, it's less useful than you might think.


Original done in Suno.



Summer Radio Edit done in Google Flow.

[wagtunes]

Ok, so, I created a "version" of the song above in google's flow app. Obviously this is not exactly the same. It was primarily a comparison of process and sound. Flow still has annoying limitations. The model tries very hard to force a three minute pop song length, which kept bumping into the repetition that I used intentionally in the previous version. There are ways around it, but, they are awkward. You can download stems, but, for me, there was an error in downloading the bass. They are broken into just four files whereas suno breaks them into eight. I called it a day and generated the Summer Radio Edit. There are key differences in instrument quality between the two. I'm not going to do this normally, it's not really my thing. In this case wags asked about process and so some of you may find this comparison instructive. My money, long term, is on flow being the better tool. For me, it is essentially free as I pay for gemini pro for other reasons. YMMV and all that.

The key difference is that with flow you are interacting with a language model that will make changes to the style prompts and guide you how to use the tool. One key aspect that helps to make changes while keeping a similar output is that you can control the seed. However, note that if you make changes to the lyrics that it regenerates the song and it finds a different path through the latent space, so, it's less useful than you might think.


Original done in Suno.



Summer Radio Edit done in Google Flow.

Thanks man. I might give this a try.

[guitarzan]

Ok, so, I created a "version" of the song above in google's flow app. Obviously this is not exactly the same. It was primarily a comparison of process and sound. Flow still has annoying limitations. The model tries very hard to force a three minute pop song length, which kept bumping into the repetition that I used intentionally in the previous version. There are ways around it, but, they are awkward. You can download stems, but, for me, there was an error in downloading the bass. They are broken into just four files whereas suno breaks them into eight. I called it a day and generated the Summer Radio Edit. There are key differences in instrument quality between the two. I'm not going to do this normally, it's not really my thing. In this case wags asked about process and so some of you may find this comparison instructive. My money, long term, is on flow being the better tool. For me, it is essentially free as I pay for gemini pro for other reasons. YMMV and all that.

The key difference is that with flow you are interacting with a language model that will make changes to the style prompts and guide you how to use the tool. One key aspect that helps to make changes while keeping a similar output is that you can control the seed. However, note that if you make changes to the lyrics that it regenerates the song and it finds a different path through the latent space, so, it's less useful than you might think.


Original done in Suno.



Summer Radio Edit done in Google Flow.

Flow’s ability to retain the original seed is great for generating new versions of a tune where everything all lines up part-wise, but Flow can’t listen to input like Suno, or at least the last time I checked several months ago…

… I used a 30 second clip generated by the Beta Google/Lyria period that I wanted to expand… Suno could clearly fully digest that Lyria clip and expand it while maintaining the actual main elements of the source tune (main riffs, beat, style, etc), where Flow (still called Producer AI at that point) was clearly just doing a crude beat detection and hallucinating all other detail… insisting that it had identified the funk heart of the tune (it was a jazz tune), etc … really not able to expand on its own original 30 second clip at all, or even identify the key or anything.

If either app could both digest the input as well as Suno and retain the seed like Flow then iterative workflow would be complete.

I preferred the Google Lyria (Producer AI/Flow) instrumental output from the beginning though… it seems to better understand how to sound like it’s jamming, putting some struggle into the performance that just isn’t even considered elsewhere. The performances seem more dynamic I suppose… small expressive things like amp sag, fingerboard noise, or overblowing on wind instruments, etc.

[TechHaus]

I think I remember BONES saying he hated Kraftwerk, but found an Australian band on a human dj'ed radio station last night that I am posting in tribute to him.

[guitarzan]

Still sounds fairly realistic to me:

https://www.flowmusic.app/song/081c2f3e ... 959c02007c

The background is not that bad. But the lead is. It sounds very synthetic, even worse than mediocre sample libraries.

With the bluegrass inspiration that was prompted, every instrument except bass takes a turn at lead, so I’m not sure which part you are referring to, but nothing sounds synthetic to me specifically… the slightly muted quality of the flat picked guitar break is expected in this genre. I was especially pleased with some of the non-quantized timing decisions and slightly mis-fingered renderings… also expected in this genre. The mandolin performance seems flawless to me. The voicing of the resonator slide guitar wouldn’t be my choice, the way it focuses on the upper mid fundamental frequencies and not on the cone resonance, but it is in keeping with the acoustic folk rock genre I called out in the prompt, bringing the Nitty Gritty Dirt Band to mind.