Claude Composer
josh.ing82 points by coloneltcb 2 days ago
82 points by coloneltcb 2 days ago
> Recently I was listening to music and doing some late night vibe coding when I had an idea. I love art and music, but unfortunately have no artistic talent whatsoever. So I wondered, maybe Claude Code does?
Do I need to read further? Seriously, everyone has talent. If you're not reaady to create things, just don't do it at all. Claude will not help you here. Be prepared to spend >400 hrs on just fiddling around, and be prepared to fail a lot. There is no shortcut.
I've got to come to the OPs defense as well. This was a remarkable demonstration of Claude performing a task thats probably very out of distribution. This would not be interesting if it were a music generation model or program, it's interesting because this is not what Claude code was explicitly trained for. The fact that it generated waveforms from scratch and built up from there is really amazing. Your cynicism was applied before even reading the article.
Author of the article here. Appreciate your sentiment here, but my goal wasn’t trying to make a hit song or shortcut the obvious very significant time and effort that goes into creating any sort of art. It was meant as a fun experiment to try to highlight a feeling that we’re barely scratching the surface of the breadth of things that agentic coding may be able to tackle. I’ve been learning guitar and taking painting classes in my free time, but it’s not my profession nor something I was encouraged to do when I was young. Thanks for the comment, it’s helpful to see ways I can improve my writing style
Yeah, it's just weird to expect people to find AI-generated art interesting when the person generating it has no unique take or talent. This is the worst case where there is absolutely 0 creativity in the process and the created "art" reflects that imo.
I don't find it interesting in an artistic way, but I do find it very interesting from an "AI experiment" angle.
I don't get what the "AI experiment" angle here is? The fact that AI can write python code that makes sounds? And if the end product isn't interesting or artistically worthwhile, what is the point?
I have a deep background in music and I think that while the creation was super basic, the way the output was so unconstrained (written by a model fine-tuned for coding), is really interesting. Listen to that last one and tell me it couldn't belong on some tv show. I've had always issues with any ai generated music because of the constraints and the way the output is so derivative. This was different to me.
At least it wrote a song, instead of stably-diffusing static into entire tracks from its training data. I can take those uninteresting notes, plug them into a DAW and build something worthwhile. I can only do this with Suno-generated stems after much faffing about with transposing and fixing rhythms, because Suno doesn't know how to write music, it just creates waveforms.
AI tools are decent at helping with code because they're editing language in a context. AI tools are terrible at helping with art because they are operating on the entirely wrong abstraction layer (in this case, waveforms) instead of the languages humans use to create art, and it's just supremely difficult to add to the context without destroying it.
What's the point if human-made art isn't interesting or artistically worthwhile?
(Most of it isn't.)
Art is on a sliding scale from "Fun study and experiment for the sake of it" to "Expresses something personal" to "Expresses something collective" to "A cultural landmark that invents a completely new expressive language, emotionally and technically."
All of those options are creatively worthwhile. Or maybe none of them are.
Take your pick.
> What's the point if human-made art isn't interesting or artistically worthwhile?
Because it is a human making it, expressing something is always worthwhile to the individual on a personal level. Even if its not "artisticallly worthwhile", the process is rewarding to the participant at the very least. Which is why a lot of people just find enjoyment in creating art even if its not commercially succesful.
But in this case, the criteria changes for the final product (the music being produced). It is not artistically worthwhile to anyone, not even the creator.
So no, a person with no talent (self claim) using an LLM to create art is much less worthwhile than a human being with no/any talent creating art on their own at all times by default.
>Even if its not "artisticallly worthwhile", the process is rewarding to the participant at the very least
I think that's the point though. What op did was rewarding to themselves, and I found it more enjoyable than a lot of music I've heard that was made by humans. So don't be a gatekeeper on enjoyment.
I've gotta come to OPs defense here. In the age of Suno indistinguishable-from-human-quality hits, this whole endeavor was an art piece and more interesting than most human OR AI music I've heard in the past year.
The medium was using the "wrong" tool for the job, which creative musicians do on a regular basis. And the output was so cool, it really felt like a relic from a different era even though it's hyper-modern.
As a musician, I find there are a lot of obsessions one can succumb to that can lead to mastery. There are those who are obsessed with the body and it’s perfect positioning and movement those who indulge in extensive experimentation (trying everything), those who listen to absolutely everything, those who meditatively repeat and repeat (sometimes at glacial tempi), and then there are those who collect every musical idea they can and gift listeners with only the best treasures they encounter. AI musicianship, if it is something that can be mastered, probably would rely on some combination of the above. It’s going to suck for a while, certainly, as there hasn’t been time for someone to put in their 10000 hours
I can't believe AI music hasn't hit the mainstream yet. It's the most amazing thing I've seen since my original ChatGPT 3.5 wtf experience. https://suno.com/playlist/fe6b642c-f4a8-4402-b775-806348640e...
This song was generated from my 2-sentence prompt about a botched trash pickup: https://suno.com/s/Bdo9jzngQ4rvQko9
music is a form of storytelling and expression of human emotions, i can't connect with some generated noise or go to its live gig
Because people don’t want to listen to robots. There was a radio station here in Norway caught playing AI music to save on royalties, it was not good for them.
The concept of AI music is extremely polarizing. One friend I played it for got visibly angry. Oddly none of the anti-‘s were musicians themselves.
Perhaps it's your sphere; I know many musicians (mostly Jazz and people in punk bands) and they aren't thrilled to say the least. Like most things, it's contextual.
> Oddly none of the anti-‘s were musicians themselves.
It is clearly plain to anyone who is a musician or hangs out with a lot of musicians that the independent music world is livid about this stuff. Everyone I’ve talked to, from acoustic songwriters to metal singers to circuit-bending pedalheads are united in their absolute hatred of this technology.
(Yes, follow-up commenter, I’ve seen the Timbaland interview)
As an independent musician, I for one welcome our AI overlords. They should not be worried about any technology that needs a human to make it remotely interesting.
They should not be worried if they aren't generic sounding independent musicians already.
Lastly, and a historical case in point, this whole conversation is a repetition of the anti-Sampler movement of the 80s and 90s. Look what that techno-leap brought us.
A new technology brings new sounds, if we all stopped treating a megalithic search engine as a personality, we'll move forwards with a lot less drama.
AI music or AI-produced music sounds a bit boring and too-perfect, like auto-tune on steroids.
Needs significant human involvement to make it interesting.
How long that remains true is another question…
AI music on a radio station doesn't really make sense. The point is to be able to create custom music tuned to your own tastes -- music that's specifically for you.
If you can generate a song with a two sentence prompt, so can anyone else. Music and art is only interesting when there’s originality or a point of view being expressed.
I really think art (as in art that’s made for it’s own sake, as opposed to jazzing up a PowerPoint slide or whatever) is by definition something AI will not make inroads in
It definitely has. Spotify regularly recommends AI artists.
Stuff like this - https://www.nme.com/news/music/ai-generated-country-track-wa...
https://www.cbsnews.com/news/meet-the-woman-behind-chart-top...
Etc
Music is about the human experience, emotions, mistakes, accidents, discoveries.
I could listen to music by real people being vulnerable and expressing themselves, or I could listen to a computer soullessly regurgitating a stock "blues" melody with inane lyrics about a trash can. Why would I ever pick the latter?
> Music is about the human experience, emotions, mistakes, accidents, discoveries.
Pro musician here.
There's piles upon piles of human-generated music soullessly regurgitating stock patterns with inane lyrics since long before Alan Turing was even flown in by the stork. Most recent popular music by far is bland sausage factory production.
Why not allow yourself to be moved by beautiful music, wheter it's machine generated or not?
Respectfully, 25 years ago someone might've said the same thing about you spending any time online at all. Today, people spend far more time on all number of "artificial" experiences. I'm not going to try to convince you that it's good or lasting or even personally entertaining to me, but it seems that it's entertaining to someone.
Art can be about that, but if it was only about that for everyone, we would never have had Britney Spears autotuned so hard that my mother assumed she was already listening to a soulless computer when she first heard Spears in c. 2000.
Me, I see the patterns too fast to even care for a second play of recorded music from a real human, only theme songs for nostalgia-inducing shows have enough of an emotional kick to get past that.
GenAI music has all the same problems as GenAI images (try asking Suno for "Just fox noises" to see what happens out of distribution), but collectively it has at least been a bit harder for me to spot the pattern behind them in aggregate, even if each song by itself still has the same problem for me as any other recording.
funny, i feel the same about most pop music from the last few decades. human created repetitive junk without originality.
These songs sound like royalty-free stock music at best. Bland and inoffensive, with the same uncanny and compressed quality that AI-generated images have too.
Borderline acceptable for elevator music is a long way from the paradigm shift you claim it is.
There are a lot of much better examples on YouTube. They only hit the uncanny valley when vocals with insipid lyrics are included.
From the OP:
> For complex AI generated music, tools like Suno and Udio are obviously in a different league as they're trained specifically on audio and can produce genuinely impressive results. But that's not what this experiment was about.
it's not art (for humans) if it's not made by a human with a human story. AI can be used as the tool with which art is made, but not as the maker itself. now, on the other hand, maybe AI can make it's own form of art for other AI's to consume. However, for the human, the creation of art will always need the human taste and story involved
You can replace the story with log output, though:
https://youtube.com/watch?v=atcqMWqB3hw
From the author:
> The instrumental and vocals were both generated using Suno with a lot fiddling around with the prompts. The video was edited by a human in kdenlive :-)
telling the AI what kind of music i want is not a story? what if i write the lyrics and use AI to add music? or if i use AI to create a song but then play and sing the song with my own instruments and voice? is reciting music not also art?
what about DJing? all DJs are doing is replaying someone elses music and recombining it in creative ways. and that is considered art. isn't that similar to telling an AI what melodies or songs to use?
i'd agree that a fully AI generated song without any human input is not art, but i would not completely reject AI use either. there is a middle ground somewhere, where that is depends on the intention of the creator.
I wouldn't be surprised if it has, or is currently in the process of, doing so. The results are good enough at this point that I think you could probably drop a few songs into a popular Spotify playlist and someone who didn't listen too closely would be fooled. I assume someone is already doing this.
It has hit the mainstream imo. Most people are content with the amount of music already available.
I agree. It's shockingly good.
It's not just good at producing complete songs though, AI has made it trivial to take garbage and make it sound good.
I largely stopped making music because imo unless you're in the top 5% of musicians AI is probably able to write better music than you.
I guess it's the same with visual artists. Unless you're really, really good it's hard to understand why anyone would produce art by hand these days.
We make art because humans are compelled to express themselves. That's it. That's the whole thing. It's not stack ranked. Humans make art because, in the words of Pile, "I want answers to some questions that I can’t speak."
The idea that you'd stop trying to express yourself because you're comparing your own artistic voice to the output of an LLM and somehow seeing it as less valid, or less worthwhile, is just sad.
I don't mean that as an insult, I mean it's genuinely sad for you and for all of us as a species.
If the reason you were making music wasn't that you enjoyed making music, perhaps stopping is the right choice for you. If that was the reason, then AI is irrelevant.
I do enjoy making music, and I don't do it "by hand". I use lots of tools (instruments, electronics, a computer for recording and mixing, the internet for distribution). As long as I'm the one directing the tools, it's still art and it's still my music.
While I'm not against AI music, do not you think there's a difference between laying down some beats in ableton with your own bass + guitar writing+playing, vs prompting an LLM?
Are you prompting the LLM in an interesting way?
Brian Eno set up a bunch of tape loops of different lengths with a few notes each and let them run until something interesting happened.
Washed Out slowed down some Italo disco and sang over it (Feel It All Around aka theme to Portlandia). Does that count? [0]
Artists gonna art.
[0] https://www.tiktok.com/@nardinyouryard/video/759472477690464...
Yes, because you have your own style. You do things your own unique way. That's what makes your music your music.
There were always musicians who were better than you. If that didn't stop you, why did AI? Were you only making music to be the best? Surely you knew that was extraordinarily unlikely. If you like making music, then make music and like it.
> I largely stopped making music because imo unless you're in the top 5% of musicians AI is probably able to write better music than you.
It won't be long before this becomes:
> I largely stopped making _____ because imo unless you're in the top 5% of making _____ AI is probably able to make _____ better than you.
Especially where _____ is anything that can be created digitally.
While the author explicitly wanted Claude to be in the creative lead here, I recently also thought about how LLMs could mirror their coding abilities in music production workflows, leaving the human as the composer and the LLM as the tool-caller.
Especially with Ableton and something like ableton-mcp-extended[1] this can go quite far. After adapting it a bit to use less tokens for tool call outputs I could get decent performance on a local model to tell me what the current device settings on a given track were. Imagine this with a more powerful machine and things like "make the lead less harsh" or "make the bass bounce" set off a chain of automatically added devices with new and interesting parameter combinations to adjust to your taste.
In a way this becomes a bit like the inspiration-inducing setting of listening to a song which is playing in another room with closed doors: by being muffled, certain aspects of the track get highlighted which normally wouldn’t be perceived as prominently.
I’ll take this opportunity to plug a couple of experiments I’ve not progressed any further but thought were fun:
- Using Claude as a “pair producer” in Ableton by giving it access to the Ableton remote script API so it can create patterns - this was 1 year ago so I’d be interested to see how newer models can do https://youtu.be/2WxSB75U6vg
- A Claude Code skill which teaches it how to arrange Ableton loops into songs (by modifying the XML as there isn’t an API for this): https://youtu.be/P6Zw6f6CEbI and https://youtu.be/tVZigxFceUE
Related: ChatGPT Canvas apps can send/receive MIDI in desktop Chrome. A little easter egg. You can use it to quickly whip up an app that controls GarageBand or Ableton or your op-1 or whatever.
It can also just make sounds with tone.js directly.
No, I definitely see why people hate on AI music. I appreciate that you had fun, but these songs suuuuuck.
Claude is excellent at a few things, decent at quite a few more. Art and music are not one of these things.
Ar they good as tools to aid in the creative process if you know how to use them and have some restraint? Oh absolutely. As replacements for actual art? Oh absolutely not.
Same goes for the entire genre of tools.
I like how the author shared the prompt + conversation transcripts. I wish OAI / Anthropic would do that when they share content demos.
Great idea. After telling claude to "make a song" and pointing it to the strudel docs, it gave me this:
https://strudel.cc/#Ly8gImFtYmVyIGRyaWZ0IiBAYnkgYW1wCi8vIEB2ZXJzaW9uIDEuMApzZXRjcHMoLjU1KQoKJDogbihgPAogIFswIH4gMiB%2BXSBbNCB%2BIDMgfl0gWzIgfiAwIH5dIFs0IDMgMiB%2BXQogIFswIH4gMiB%2BXSBbNCB%2BIDUgfl0gWzMgfiAyIH5dIFswIH4gfiB%2BXQo%2BKjJgKS5zY2FsZSgiRDQ6bWlub3I6cGVudGF0b25pYyIpCi5zb3VuZCgiZ21fZWxlY3RyaWNfZ3VpdGFyX211dGVkIikKLmxwZihzaW5lLnJhbmdlKDYwMCwgMjIwMCkuc2xvdygxNikpLmxwcSgzKQouZGVsYXkoIi40Oi4xMjU6LjYiKS5yb29tKC42KS5nYWluKC43KQoub2ZmKDEvMTYsIHggPT4geC5hZGQoNykuZ2FpbiguMykuZGVsYXkoLjUpKQoKJDogbigiPFswIDRdIFszIDJdIFswIDVdIFszIH5dPi8yIikKLnNjYWxlKCJEMjptaW5vcjpwZW50YXRvbmljIikKLnNvdW5kKCJzYXd0b290aCx0cmlhbmdsZSIpLmxwZig1MDApCi5hZHNyKCIuMDE6LjI6LjY6LjQiKS5nYWluKC43NSkKCiQ6IG4oIjxbMCw0LDddIFsyLDUsOV0gWzAsMyw3XSBbMiw0LDldPi80IikKLnNjYWxlKCJEMzptaW5vciIpCi5zb3VuZCgiZ21fZXBpYW5vMToxIikKLnJvb20oLjgpLmdhaW4oLjM1KS5kZWxheSgiLjM6LjI1Oi41IikKLmxwZihzaW5lLnJhbmdlKDgwMCwgMzAwMCkuc2xvdygzMikpCgokOiBzdGFjaygKICBzKCJiZCIpLnN0cnVjdCgiW3ggfiB%2BIH5dIFt4IH4gfiB%2BXSIpLAogIHMoIn4gW3JpbToxLCBzZDoyXSIpLmdhaW4oLjgpLAogIHMoImhoKjgiKS5nYWluKHNhdy5yYW5nZSguMiwgLjcpKS5wYW4oc2luZSksCiAgcygifiB%2BIH4gb2giKS5tYXNrKCI8MCAwIDAgMT4vOCIpLmdhaW4oLjQpCikuYmFuaygiUm9sYW5kVFI4MDgiKQoucm9vbSguMykuc2hhcGUoLjIpCgokOiBuKCJ%2BIDxbfiA2XSBbNCB%2BXSBbfiAzXSBbNSB%2BXT4iKQouc2NhbGUoIkQ1Om1pbm9yOnBlbnRhdG9uaWMiKQouc291bmQoImdtX211c2ljX2JveCIpCi5yb29tKC45KS5kZWxheSgiLjU6LjE2Oi43IikKLmdhaW4ocGVybGluLnJhbmdlKC4zLCAuNikpCi5wYW4oc2luZS5zbG93KDgpKQ%3D%3D
It layers a pentatonic guitar melody with filter sweep, a saw/triangle bass, warm e-piano chords, TR-808 drums, and a sparse music box that drifts across the stereo field.
I do acknowledge the possiblity that it might be heavily plagiarized from an original composition in the training set - I wouldn't know.
Curious to see how this worked, I tried this on Deepseek using Claude Code Router, following the author’s guide, with two small changes: Make it an emo song that uses acoustic guitar (or, obviously an equivalent), and it could install one text-to-speech tool using Python.
It double-tracked the vocals like freaking Elliott Smith, which cracked me up.
Very interesting experiment! I tried something related half a year ago (LLMs writing midi files, musical notation or guitar tabs), but directly creating audio with Python and sine waves is a pretty original approach.
>I love art and music, but unfortunately have no artistic talent whatsoever.
Then go pay someone to teach you to play <instrument>, and you'll get a life skill that will be satisfying to watch grow, instead of whatever this soulless crap is.
edit: Oh god after listening to those samples, send Claude to the same music teacher you choose...
Oh man I love this so much. The prompts made me laugh so hard. Great experiment.
We alrrady had Cursor Composer last year, so it sounds like a step back.
Making music with AI is a new hobby of mine.
My journey started after my wife found a Ukulele on the side of the road near where I lived a few years ago and took it home. Then often when I had a short break, I started just tugging at strings, trying to fully internalize the sound of each note and how they relate... After a few months, I learned about Suno and I started uploading short tunes and made full songs out of them. I basically produced a couple of new songs each week and my Ukulele playing got a lot better and I can now do custom chords. I'm all self taught so I literally don't know any of the formal rules of music. I shun all the theory about chords and composition like chorus, bridge, outro... I just give the AI the full text and so long as the main tune is repeated enough times with appropriate variations, I'm fine with it.
TBH, as a software engineer, I was a bit surprised at how rigid music is. Isn't it supposed to be creative? Rules stand in the way of that. I try to focus purely on what sounds good. For me, even the lyrics are just about the sound of the voice, I don't really care what they say, so long as it makes a vague general statement (with multiple interpretations) and not cheesy in any way.
_Neon Dreams_ is ELO × Daft Punk.
I'm not okay with AI taking the place of human creativity