Music-making in our AI obsessed world

With this, I was trying just to go for the most wild mashup of styles/genres I could. I gave it the sample of glitch techno, told it to write lyrics as "an alt country ballad" and the style was "hard rock grunge".

The first attempt was very very "generic Suno". However, when I dialed up the audio influence tab, this is what I got.

Of course, the weakest link is the lyrics, and the fact that the heavy metal guitars sound pretty cool made me think "hey actually I might be able to work with this!". To me, that's the ultimate fun of Suno.

And after sitting with it for like a half hour, I couldn't resist going back, writing a full set of my own lyrics, iterating etc as I always do. The result is "Burning Lava".

Just another random thing that popped into my head. I thought of it as a kind of "Austin Powers" esque lounge/jazz/psych theme song for a fictional action hero (that has no name and no description). Then, like I do with many of my Suno songs, I reimagined it as breakbeat. Because always breakbeat.

I think this is another example of Suno just totally taking some random snippet of music of mine directly out of my head and making something real out of it. It's just fun.

I was trying to get strange sounds out of a synth, and then I put them into Suno basically (isn't that how most of these start?).

Though like most of my Suno creations, I had a pretty good vision for what I wanted it to make. This sample sounded kind of like galloping horses, so I thought "D&D fantasy music". Then I got dollar signs in my eyes and went grazy and made, what, like 8 of them? I thought to myself, DM (Dungeon Masters, the ones who run D&D games) are always looking for music for the background for their games right? I could just generate a gazillion hours of this stuff and put it up on YouTube and retire!

Well two problems. The most obvious is that if I can do this, anyone else can, and I'm sure that YouTube is already innundated with it. Secondly, I got carried away and put way too much effort into coming up with cool ideas for the samples.

As it went on, I realized that I didn't need to create a new Ableton project for each one of these. I didn't even have to "bounce in place" and drag the WAV to the desktop, then the suno window. I already had IAC set up for my sequencer experiments, so I just had Suno record directly from Ableton. Lol.

This one was an experiment in how much I felt like writing on my own before putting it into Suno. I actually set out to write the song completely in Ableton, myself, the "traditional way". After I got the first verse in chorus, I immediately caved and put it into Suno. At that point, I had it sounding pretty good, and it was easy and fun to write the rest of the song. So the full sequence was Ableton → write chords/lyridcs/melody → Suno to figure out how it sounds → Finish writing it.

Now, some people have suggested to me: why don't you go back and re-record the Suno version, to truly make it your own?

I don't really see the point. If I had finished recording my Ableton version, I would have had a demo. Once I put it into Suno, I had a demo. If I re-recorded it myself I would still have a demo, and it would be organic, but I would have spent 10 hours for something that is only marginally more useful (I didn't say better!) than what I currently have.

I was actually playing my actual acoustic guitar and I found a cool chord riff that I liked. I quickly recorded it and uploaded it to Suno to see what I could do with it.

This is not that sample.

The original sample got flagged as copyright encumbered. When I asked Claude what song it could possibly be (I couldn't recall myself), it suggested that it was just too generic of a sample, and that Suno was being conservative. This makes sense, I've heard of Suno rejecting people's lyrics because they were like "I love you like the stars in the sky". Sorry, someone has used that already!

I recorded a different chord progression and then wrote these lyrics. I think it came out pretty good, but I was hoping for a more MGMT/Tame Impala vibe, and this sounds too much like Creed.

Started with just the beat. I originally just specified the style and prompted suno with the chorus, with markers for [Verse 1 here] and [Verse 2, rapping] there. That didn't work too well, so then I just took inspiration from the basic melody and rhythm that it had produced and wrote my own complete lyrics (including the rap!).

A friend got engaged today! In honor of the occassion, I guess I decided to write a sappy, sappy love song. Started with a vocal sample (not pictured).

This is a full song, that I picked from about 8 candidates. I provided the guitar sample, and some prompting to get the right style. Then I wrote half the lyrics to see if it was coming out ok. It was, so I wrote the rest of the lyrics.

I was actually considering taking this further, recording a version where I play guitar and sing it myself. I "learned" the chords to....my own song....and was ready to perform, but my mic cable crapped out. I'm still planning to do this and will update.

This is the perfect application of Suno, honestly. I just wrote some quick lyrics to illustrate the point of a new website/pamphlet I made, https://nevergivethemyourface.com. [Side note: I wanted that site to be anonymous but I guess I'm violating that here. I'm sure someone would have been able to figure it out]

I knew the feel of the song I wanted, Dropkick Murphys/heartland punk. So I just wrote it up and typed it into Suno.

The point is the lyrics, the artistic statement, not the recording or the particular expression. It would not have made sense for me to spend 5-8 hours recording and mixing this, for a 90 second song that 8 people are going to listen to.

Once again, it is a Saturday morning and I am singing a song that popped into my head (er, I mean....I wrote) to my cat.

As usual, I sing it lazily into my phone and send to my wife and her sister. But this time! I decided to take the sample and turn it into a Suno song, for max camp and comedic effect.

And before I even did that, I made an exercise out of it. I brought the original sample into Live and added a verse, plus some vocal harmonies. Still just really a layered vocal sketch and I didn't put much effort into getting this on time or even in tune.

Finally I decided to completely nuts and recorded a few more layers for Sample+. That one is kind of unhinged and I feel a little bit embarassed posting it. I already had the target genre in mind when I added that stuff.

This song was the end result of hours and hours of arranging and mixing. It still counts as a Suno experiment, and I'm not claiming otherwsise. But it is a hybrid, I'm also claiming it as a song I wrote. I am posting it on my original demos website.

The methodology started with humming a riff again. I tried really hard to get Suno to do "Only Guitar". I had an idea that I could hum all the parts individually and then mash them together in Ableton. That basically didn't work at all. Suno would always produce bass and drums. So I adjusted course.

I extracted "Stems" from the 50 second song it produced and got a "guitar stem". Though it wasn't a crisp guitar stem, there was all kinds of bleed and effects and layering. I think the idea that Suno can produce stems gets people excited, but isn't always particularly useful.

Anyways, realizing how messy this sample was, I thought...."chop it up!". I just resequenced it heavily and wrote a new song around it.

So in the final analysis, Suno worked as kind of a synth, where I hummed something and got a guitar sample.

I thought, what if I take a chiptune and make it jazz?

Another exercise in taking a short clip of my own and just slapping a random style prompt and some lyrics on it. Suno can generate these faster than I can dream them up, which is par for the course for genAI I guess.

At this point, I'm generating a lot of these. Of course none of them are particularly notable or interesting really. That's AI slop for you, right? But I still want to post, so here are a bunch of different tracks from this week.

One thing I've noticed is that turning a 20 second idea into a 3 minute song? Yeah it doesn't really work. But it makes me think, could I splice a few different Suno versions together somehow?

This is it. I took the full mastered version of my song You're A Star, the version that's you know on Spotify and YouTube Music (where it has a beautiful 37 views), and plugged it into fucking Suno. I watched as my creative life of the past 20+ years melted into a puddle before my eyes. I knew it would come to this, I knew this was the final test, the final boss as it were. I actually did this a few days ago and resisted posting it.

In a sense, this version is just as much a demo as my version. Does the Suno version sound better? I would say it just sounds more polished in a way, maybe more exaggerated? More full? I don't know. I don't know anything. I certainly don't know what to do.

This is the first one I'm actually disapointed with. I got ambitious and tried to write a song with a varied and ridiculous description of the "style". This is because I saw Suno extracting things like this from my previous samples, like "percussive beatboxing followed by bass melody", etc. I actually wrote the following from scratch:

Album oriented classic rock song, pomp rock, prog rock. D minor, 118 BPM. Intracate fingerstyle guitar part opening, swells into percussive distortion isolated with no other instruments, orchestral strings and hammond organ are added and swell to conclude in an a cappella intro hook, intro hook in 4 part male harmony, no instruments. Distorted guitar re-enters along with 2nd guitar playing bluesy hooks. Verse 1. Double time chorus with big band horns for Chorus 1. Fingerstyle guitar reprises. Immediate Chorus 2 at double time. Hook heavy bridge/instrumental interlude in 7/4 time, promiintent bass line. Complete style change for verse 2, lo fi electronic breakbeat drums, multiple build ups. Chorus 3, normal time, same instrumentation as previous choruses. Guitar solo. Chorus 4, same as Chorus 3. Outro, same chord progression as Chorus 3 and 4, with fingerstyle guitar slowly re-entering, becoming the only instrument, then the whole thing fading out.

And to be even more ambitious, I wrote the lyrics that follow from this too, before I ever clicked generate. I went through about 12 iterations on this 5 minute monstrosity, before I just called it good enough. (I actually took it into Ableton to try and save it, but nothing that I couldn't have ended up doing in the Suno editor).

I'm also starting to notice that the engine really likes to embellish and just plain noodle a lot. You're not going to get a "tight three minutes" from Suno.

I.....I'm sorry.

Same "stuck in my head just hum into the mic" process here. I think the workflow I'm coming around to is successive iterations.

• Don't bother writing lyrics until the style prompt results in a genre/feel that fits your vision for the song.
• Don't write too many lyrics if the singing is off, try to rectify it.
• Listen to the songs, immediately disqualify ones that start wrong or have awkard melodies.
• Don't listen too much though, because your ear goes dead to the subtleties. I've even done some "palate cleansing" and listened to completely different genres in between checking versions.

Then it's just putting on finishing touches like adding bridges and repeating choruses the right number of times. Violin!

He was on my lap. I started singing into the mic on my desk. I corrected the auto detected lyric ("Shuck-a-muck itty that"). Then for the styles I typed....."salsa". Just that.

This was the workflow. I started humming a simple rap-rock beatbox to myself. Then I thought, wait, Suno has in browser recording of samples. Why the fuck not? .…I literally ended up spending as much time uploading and writing this description as I did "making" the tracks. Actually I spent about 10 minutes just listening to this shit, and refining, and it getting better and better. I think I might look back at this as my "Wow, I hope other people don't find out about this!" moment.

In 2022, my friend Richard asked me to make him a motivational song that he could play when lifting weights. I did it. I doubt he ever used it but I did it. I uploaded it to Suno and asked for a remix, using a lot of the extracted prompt, and adding '80s aerobics" and "over the top".

My very first experiment with this workflow and it's just insane. I literally uploaded a nine second clip of a basic-ish arp pattern with a little drum pattern I programmed in. Click the "Suno" tab to see what I got out! Also I named the song Sample because that's what I named the original....sample. And the name stuck 🙃

So soon after I made "Skin", I was just listening to some actual human folk music. And an alternate/similar lyric to something they sang popped into my head. Of course I headed right to Suno. I typed in the lyrics that were on my mind, and then just kind of went with it. Gave it a triple cheeseburger with bacon level list of genres and instruments and this is what we got.

I wrote all the lyrics, but mostly I wanted to hear what Suno would come up with for a future electro blues glitchy thing. I am very impressed with this, I think it's extremely unique and interesting. Yes, for anyone who has done text to song, there's kind of a magic about it, that your idea became real. It's especially potent when you wrote the lyrics yourself though.