No, I am not a Nazi looking for help among the French, but rather a lyricist seeking a musical composer to work with. I’m not a performance-level guitarist, but I’m skilled enough to write out chords and melodies; my musical skills, however, are nowhere near professional enough to produce all the sounds polished songs require. So how do I go about soliciting a song-writing partner?
I’m certainly willing to have him/her take half credit, and even to rewrite my chords altogether and set a different melody to a tune of mine. That’s not the issue I face. The issue is finding someone simpatico who is inventive and looking for a lyricist.
I’ve tried working with a couple of musician friends but it’s never really worked out. One close friend can’t even listen to my drafts because he stops me 30 seconds in to critique my guitar-playing, which really doesn’t get the point of what I’m trying to show him. That same friend also refuses –on principle, he says—to listen to the versions I’ve gotten an AI program to make for me. “I hate AI,” he says, and then proceeds to denounce the soul-sucking soulless monstrosity AI presents to all artists everywhere.
I think the AI program’s versions are pretty good actually, sometimes better than “pretty good,” but I find Ai tough to really collaborate with—it ignores some of my suggestions (when I prompt it to include a saxophone in the instrumentation, for example, it gives me a sax-less version, even after I repeat my request) and I’m not sure how the rights to an AI collaboration would work. I’d rather work with a person in the same room, at least on occasion. I’ve got another friend, a professional musician, who lives in my hometown, but unfortunately that hometown is 1500 miles from my current address. We see each other every other year or so, so that hasn’t worked out either.
So what do I do? Is there some online venue I could advertise on for what I want? Some IRL venues for me to hang out in? Maybe there’s someone on the Straight Dope who fits my bill? I can include some of the AI clips to give an idea of the sort of lyrics I write and the music I’ve come up with on my own and with AI’s help.
I’m using the AI mostly as a way to give potential collaborators some idea of what the song could sound like, with decent instrumentation, and a good singing voice. AI makes some mistakes, which I can explain away: It mispronounces some words, emphasizes others on the wrong syllable, and makes typical AI mistakes–repeating lines I don’t want repeated, sometimes it provides weird effects (crowd noise, for example). But it sings better than I can, and plays many more instruments than I do.
I think you should work on your clips/examples/samples first, long before your many written words. Make a demo track to present (edit, to another musician, like me) & describe what you imagine added to it.
The friend you already approached was probably your best-case audience and potential collaborator candidate and he didn’t last half a minute.
That seems rather short sighted. Some of the best Beatles songs started out as vague sketches on John’s guitar!
I haven’t had much luck trying to use AI as a writing aid. Which AI have you been trying? With ChatGPT at least, it seems to struggle to even complete a song. Though it comes up with the occasional fairly interesting lyric line.
That is probably ideal, but I’ve had good results working remotely with musician friends. We just start with a framework and then ship additional overdub tracks back and forth to import into our respective DAWs.
That would be a good plan. Unfortunately the board software does not allow explicit upload of audio, even though there is an icon for it. (I ran into this a while ago). So you would need to put them on some publicly accessible hosting platform.
For what it’s worth, here are a few samples of my stuff; some solo and some with co-writers.
I used the AI program called Suno. Aside from ignoring my specific instructions, it works pretty well. The free version lets you make up to 10 songs a day so I keep trying until I get a version that comes close to what I have in mind.
OMG, that thing is scary good. AI is getting dangerous. Almost makes me feel it’s time to hang up my instruments and check into a nursing home for my twilight years.
But not quite. I’m already working with a couple of friends but always open to new directions. I assume you’ve checked out my examples, so you can judge whether there is any synergy.
So, what input are you looking for from a collaborator?
Just looking to see if the AI version suggests anything musical to him/her. Even the AI comes up with different musical genres and styles, so I expect a person could come up with even more varieties.
Out of curiosity, how much prompt did you provide for the example you gave? Could be anything from just a title & descriptive phrase or two, to a complete lyric plus detailed arrangement instructions?
As I said, that site seems to be scarily capable of producing something quite convincing from very minimal prompting.
To go back to the question: what do you want from a collaborator? Someone to come up with ways for taking a partial start further (eg suggesting a bridge)? Music or lyrics? An arranger or orchestrator? Someone who can use a DAW to put it all together (engineer / producer)?
I can wear several of those hats, but I’m not quite sure what you are looking for…?
Maybe you could show an example of something you have and what you would like help with?
A Lennon to my McCartney? A McCartney to my Lennon? What I want is someone who can improve what I’ve started out with. Since I’m more inclined to start a song with lyrics, and then come up with suitable chords and melody, I probably need someone whose primary talent is musical, though there are a few melodies I’ve got that I haven’t thought of suitable lyrics for .
Thanks. Quite helpful. I’d actually caught the “you/you” rhyme myself, and I’d fed a newer version into the Suno AI already: Reluctant Farewell by SensoryPrelude0927 | Suno that incorporated a third (slightly reworked) verse as well. I agree it could use a middle 8, though I’ve written some chords that are an improvement on the Suno version.
I would like to ask: what is your overall process and end target? Do you hope to train a program like Suno into producing a ‘finished version’, or are you just using it as an idea generator?
As you will see from my examples, I mostly play everything myself (often with contributions from other musician friends) and nowadays construct the projects in a DAW (Reaper).
Suno seems to be able to produce a scarily good demo very quickly from minimal prompting. But it’s not very flexible: you might be able to chop up its output, and reorder sections, for example, but trying to add something like a bridge or middle 8 won’t work because you can’t replicate the sound and feel of the generated track.
And as far as I can see it doesn’t have any ability to export individual instrument and vocal tracks as stems? (Maybe I’ve overlooked an option, or that’s only available in the upgraded paid version)? And I’m pretty sure it can’t import a new instrument or vocal track to add to or replace part of a generated song?
I guess I’m still asking, what sort of collaboration do you want?
The way it works with some of my other friends is, one of us constructs a very basic framework (often just a vocal and acoustic guitar). And we import that into our DAWs and then ship additional overdub tracks back and forth to construct the final arrangement.
Ah. A bit more research: it CAN do that but you have to subscribe to the paid version.
To be honest, I don’t want to do that (or really, use this or any similar program). It feels too much like cheating: it’s not really ‘your’ song at all.
Sooner or later I suppose this sort of thing is going to take over (in fact I recall seeing a report of an AI-generated song being a hit on the country music charts).
But I’m not ready to give up on the human element yet….
what is your overall process and end target? Do you hope to train a program like Suno into producing a ‘finished version’, or are you just using it as an idea generator?
My end is to have a performance made of my (or our) final product). I don’t have faith in SUNO’s ability to get what I’m aiming for, but I have less faith that I will be able to produce by myself a version that will knock anyone’s socks off.
In several cases, including “Reluctant Farewell,” I think I’ve actually composed more interesting music than Suno has, but recording myself singing and playing guitar is more laborious and time-consuming a process than letting the AI do it, so I’m employing AI’s version which yields a passable product for a potential collaborator to think “I can work with this.”
I would much rather hear the original idea than something artificially ‘completed’.
As I said, some of the Beatles greatest work started as a sketch on John’s guitar.
Yes, it’s a bit time consuming. But it’s Human. Would you want to have an AI produce a Mona Lisa from just a prompt: woman with an enigmatic smile?
And practically: it’s not really that hard: does it really take that long to knock out a demo with just guitar and vocal? Have some faith in yourself and give it a go, I’d say.
You haven’t heard my voice, or my guitar playing. One is worse than the other.
I find recording myself to be very time-consuming. Multiple takes, frequent do-overs, plus setting up various machines, instruments, etc. Doing it with AI takes a few minutes, and I end up with a much more acceptable product for my purposes, which are simply to give a rough idea of what I’d like the final product to end up as.