https://samhenrycliff.medium.com/my-surprisingly-fun-ca-h-gr...
https://flippa.com/12100071 - I was wondering why SmashHaus was for sale. (no affiliation) Peak value. It's only downhill from here for outsourced music.
> My bad—I forgot to hook up the sound system.
And then it started playing jazz, which I'm not mad about. Nice to see Google trying fun stuff.
- Vol I (AdP)
- Vol II (AdP)
- Flying Microtonal Banana (KGatLW)
I just wanted to write this comment because it will be almost impenetrable to anyone who doesn't already know.
Now I'm only replying because I'll take any opportunity to prop King Gizzard and the Lizard Wizard, which is the third item of the original three dot points.
KG = KG and LW = KGatLW = King Gizzard and the Lizard Wizard.
I don't like all of KGatLW's music but, as someone who is also a big fan of Frank Zappa's extensive corpus of works, I love their versatility and their willingness to be versatile.
One of my favourite performances of KG: https://www.youtube.com/watch?v=MI_XU1iKRRc (I can't believe this is from ten years ago!)
https://www.youtube.com/watch?v=0Ssi-9wS1so
I think this is rather an AGI benchmark than a pelican.
When I first watched the AdP KEXP performance, it felt like my musical knowledge up to that point was just preparing me for AdP. I've been through micro-tonal (King Gizzard and the Lizard Wizard, specially their Flying Microtonal Banana album), polyrhythms and time changes (Tool, The Mars Volta, Metallica's St. Anger album, various other bits and pieces), looping (Party Dozen -ish-, Adam Page[0][1][2], various Math Rock bands, primarily Battles), and other general "out-there-ness" (Arthur Brown[3], Frank Zappa, Mr. Bungle[4]).
[0]: https://www.youtube.com/watch?v=95h3M6BG2QM
[1]: https://www.youtube.com/watch?v=kr1ykpNPOSg (17 years ago, holy shit)
[2]: https://www.youtube.com/watch?v=Er--olP_8wQ
[3]: https://www.youtube.com/watch?v=yXYgLPBfF_w
[4]: https://www.youtube.com/watch?v=E4YvvhKW7bA
The Adam Page stuff is all, essentially, stream of consciousness, unique to the individual performance. He had a monthly residency at a pub in the city a long while back, and I reckon I went to see him ten times over the course of a year and each performance was unique. I think they're all recorded on Bandcamp somewhere, hey, yes, here: https://adampage.bandcamp.com/music. I even named one of his songs: "Preventing a future disaster" (from September 2014, although it's misspelled). I find part of the enjoyment of looping music is in the performance itself, the combination of conducting and choreography.
I imagine that AI can be trained on Angine de Poitrine, but it would never really be able to create something like that no matter how many text prompts it was given.
I'm struggling to get a JJJ style "Like a Version" cover of Sleaford Mods doing King Stingray or vice versa .. no joy in the sense of it's a struggle to get anything that sounds anything like either band.
The text it generates midway sounds promising .. and then it plays audio that has zero of the unique elements of either.
It does explicitly state that it can't (or won't) imitate the style of a band(?).
So it's an evolutionary step in my view, rather than a revolution.
They all kinda suck, you do have to run generation many times unless you're very lucky.
I and my friend wrote 10 albums between 1997 and 2007. we went solo for geographical regions, and i stopped writing music altogether in 2017 or so, only doing arrangements, mashups, mastering.
I can't use this google product to make a song. source for credentials is my youtube, soundclick, and soundcloud accounts, e.g. https://www.youtube.com/watch?v=HXro-e0e7aA
and now i don't even know if i ever want to get back into music, because i can rapidly generate a "good enough for the moment" track, like when the south korean president tried to coup: https://soundcloud.com/djoutcold/i-aint-even-writing-music-a... Oh by the way the lyrics are in lojban except for "... and the people were pissed"
the oldest stuff on those three sites i mentioned is all hand written by me over the years.
Before the purchase, the quality of generations had been going down for a while (IMO; subjective and anecdotal). I tested multiple iterations of their chat interface and was never thrilled with its ability to actually understand or adhere to prompts. I had liked their previous (Suno/Udio-like) iteration (Riffusion).
Curious to hear how it performs for people now and whether anything has improved.
Like a strange form of Gell Mann amnesia, where all AI output is probably this bad, but if we don't know any better, we don't know just how bad it is.
various genres: https://suno.com/s/Oc5842XzzuBTk4Ma https://suno.com/s/RdmFOKpbi4zyVbRf https://suno.com/s/J4Z8t8jU9JXVJ1DB https://suno.com/s/OhfzCYkmcZhFf1Pk https://suno.com/s/VYHHLW7Hkw2uHjrb https://suno.com/s/cTu7AkoOdAyi0eWz https://suno.com/s/QvOExImOVzo1b2Gl https://suno.com/s/MASINon9lGr9JPLS https://suno.com/s/ujpTfZwVdAKy9W0h https://suno.com/s/DwekDLuEzgyNpYGQ https://suno.com/s/psWqWzDQa6Aq96Pk https://suno.com/s/JEM8G2RxD35ZUpGy
also if you like enders game lol: https://suno.com/s/gQ8eGNgnkfktl0Xq
That's the craziest shit I've ever heard.
Nope. Not for me, not ever.
I know what music I like. I said I wanted "dystopian, ambient, droning music with ear-filling, warm bass. No drums or beat."
What came out was some pretty generic dubstep that one might hear on a Verizon commercial circa 2018. Subjective, sure, but a big miss given the instructions I gave it.
Now let's say I ask it to generate code to scrape all real estate listings that were recently taken off the market. The output looks good enough to me, and I'm happy. But is the underlying architecture just as bad as the music?
If the humans of the 2020s hadn't given up their souls long before this era of AI came out, we wouldn't be discussing this ad nauseam every single day for years on HN. There's no comparison. Humans are vastly superior at music.
People need to get a grip and get out there. I know a significant majority of people reading this are into enough hobbies to know at least one instrument well or can sing. They should just do it for fun and as loudly as possible everywhere they can get away with it. We used to celebrate that sort of thing as a necessary part of maintaining a civil and thoughtful world. Music is an activity before it is a product to be sold. AI is a recording technique built from lots of recordings.
One might have a silly counterargument like "what if there are microphones everywhere stealing my work?", but not enough people consider these days that the corporate world is absolutely terrified of trying to sell what is beautiful. It was considered too risky even when they believed in it with all their heart and were deliberately trying to. What makes the most money is the average, not the exceptional. There's no good excuse people put up against pursuing music other than neurotic irrationalities that come from being chronically online.
(edited for pronouns - overuse of "you")
I’m certainly no fan of the economic structure of major labels (or any capitalist entity), but despite that they do sometimes still release some good stuff.
I edited my comment to remove all the "you"s. I have a bad habit of using it in the "fourth person", so to speak.
> hidden microphones
I meant bootlegs. There was a story here making the rounds recently about a huge archive of them, and then some discussions went off the rails about AI.
I guess to protect kids, we need to restrict them from.. checks notes music?
Pythagoras argued that music is essentially number and proportion. If beauty is found in the geometry of sound, then the "belief" of the architect is secondary to the elegance of the structure.
Have you not heard music before? Is Suno your first experience with music-shaped sounds? Because, buddy, this is wild. You're not getting Rumours out of an AI. You're not getting Time (The Revelator) out of AI. London Calling does not spring from the geometry of sound.
I still don't think you're saying anything that refutes the geometry of sound argument, however. If you heard an AI song you liked, and didnt know it was AI, and found out after the fact, would you be rational enough to accept you could be wrong? Or would it turn you off to the song irrationally?
To name one, "Saving All My Love For You" should never be played at a wedding because the song is about having an affair with a married guy with kids. But no one listens to the Lyrics. They just hear the chorus. It's a hit for other reasons, not because of lyrics.
Similarly, few people listen to the lyrics of ""Rainy Day Women #12 & 35" (everybody must get stoned). It is not about drugs.
Heck, famously there's Bush Jr. (or more likely some PR person) using "Born in the USA" as a pro-American song. It's not a pro-America song at all. I wouldn't call it anti-American but it's definitely a song entirely about problems in America. Not praise.
https://www.flowmusic.app/space/27dcebeb-7aae-4a0d-9031-308b...
This is really sweet as a tool for creating ways to create music. Its a shame you can just prompt -> finished project when its so easy for anyone to create their own music with the right tools and motivation.
The workflow feels wrong. it should be closer to a DAW with chat, where the model outputs stems, samples and arrangement parts instead of one finished track. Then you could target a specific sound, section or idea and actually develop it.
They could attempt messy stem splitting like all of the other tools have done for a few years now, but those aren't really usable in a production setting beyond small samples you were already going to chop/distort.
I am not sure if it is there yet either, but imo your UX vision for it is the correct one, so if the tech is still not quite there yet, it is just a matter of time. But the AI-powered DAW UX is imo where it will eventually end up.
One of the key issues that I encountered after a few song generations is that it feels very rushed, like it's constrained to this 3 minute limit per song so it forces every section of the song to conform to a very specific structure. I tried increasing the limit to 4 minutes but it still gave me 3 minute songs.
Honestly, I feel like this product is showing up a bit late to the party and it's not really feeling particularly innovative. There's nothing egregiously bad about it, but it doesn't seem to add anything new or special that I could notice.
I don't have a microphone hooked up so I can't try the voice interface, but it would be really fun if you could sing to it in order to iteratively compose a song. It could clean up your voice a bit and add music. Or being able to hum out a beat which it converts into a track which you slowly build up. Is anyone able to try if those capabilities are possible with the existing product?
Overall, I'm not sure if a chat interface is the best way to produce a song. It feels very restrictive to have full songs as the primary iteration mechanism. In a text file and in code you can inspect or modify different sections or components very easily. I think a more human-focused tool would provide an on-ramp towards full music production, where you can focus on the parts that you care about and enjoy, while the AI tool fills the other parts with sausage. Right now you can chat with the tool but it appears to be quite limited in the kind of changes that it can make.
"solo banjo instrumental, strictly no other instruments" ... ten seconds later: drums, a fiddle, and a guitar join in.
The music is all just very average, it sounds like the most average song with the most average chord progression/drum pattern per genre.
I guess that makes sense if these are most likely next audio token predictors... but it'd be cool if there was a way to inject some type of creativity/novelty into these, or at least tune up the temperature.
Everything so far just sounds like stock library music to me.
Another was a set of songs that helped me emotionally regulate on the drive home after couples therapy. The lyrics contained grounding exercises that helped maintain awareness and presence and contained mindfulness practices.
Both did their job, but they were also music for utility, not necessarily for artistic enjoyment. So it's not entirely an apples to apples comparison.
By the time I finish writing this comment - yours is 10 minutes old - someone will have vibe coded one, probably.
Also feels like an easy feature for someone like Suno to add, to help subscription retention.
But something like NotebookLM emphasizing subtle mnemonic devices set to music..
Also, probably someone will game an algorithm to get revenue from a bajillion tracks of lofi slop.
Slop is starting to dominate uploads to some music services, so I think it will only get worse from here
I could understand if this was an API that people built products around, but it seems to be geared directly at consumers.
How many iterations (arrangements and recordings) do you think a typical Billboard pop song goes through before it's ready for a final mix and mastering?
Go find a YouTube of someone doing this work, it is kind of mind blowing. Given how expensive studio time is, you realize why it costs so much for a popular artist to produce a polished album.
Odds are for every 200 ai songs you generate , 2 or 3 are decent.
Anyway. UMG will probably force you to sign over training rights in future record deals.
The models still can't rap. Sounds like if you asked someone who didn't know what rap was to read a script
I especially love the glitchy ui sounds, although I suspect it's hardly intentional.
Sloppy humans create sloppy output. The AI is just an amplifier, it has no motive
yes! go tell to teenager kids harmonize by typing and not editing sheet score a > 4 instrument melody without any music theory background or in-ear training
AI being "fun" for kids who aren't able to grasp the negative externalities of using AI doesn't make it something to admire
You recognize obvious slop as slop. But this is a survivor bias-like phenomenon. You have no idea what goes unnoticed.
If you're talking about stuff posted to communities that are self-selecting for AI art submissions, that's another kind of fallacy.
You. Are. Not. Creating. Anything.
You are prompting. Then tweaking, changing, adjusting, etc. The tech is incredible, don’t get me wrong, but it’s advertised so blatantly as the user doing the creating.
Use it as a creativity tool, but don’t get caught up in the false belief that what it spits out is something you created.
Old man yells at cloud. Going back to my cave now.
See for example Suno Studio, which is not very good in my opinion, but shows the direction they’re going.
The other 10% is editing, which for me involves minor color adjustments, highlights, shadows, cropping, etc. I make all the decisions.
AI can generate an image based on a prompt, and that’s fine, but I would never, never claim to have created that output myself.
I'm no photographer, but do anyone not have that Sidewinder seekerhead in the brain that gives you the blaring tone when a great composition is in front of you(including 3D off-boresight warnings)?
Does the guy who tells the composer "write a song" create?
No.
The line is somewhere in the middle
I find that people who rush to negative judgement of LLM-generated art are not going far enough in the creative process to properly judge just how much juice there is to be squeezed out of those 50-billion-dimensional spaces.
They're a music store, they sell music, both to own, but also renting their vast library out.
Google should learn not to shit where they eat.
If anything it gives Google control of the entire production->sale->delivery process.
I'm honestly not seeing a downside for Google here, can you elaborate?
The downside for Google is, ultimately, the death of the company. Nobody wants AI slop, and go out of their way to actively avoid it and punish companies that promote it. Google already is running a huge risk by pushing Gemini into every service, and permanently burning customers and users with it.
Microsoft is already seeing the downside of trying to Copilot everything. Their software is now partly slop, shit randomly breaks, companies cancel Azure/Office subscriptions and move to on-prem, FOSS, etc. They've pumped their brakes quite a lot, but the damage may be too great to mitigate now.
If Google wants to lose money in the long run, then by all means, please continue.
Once you have that particular brand of cancer, its too late to save the company without drastic measures.
Nice
The sound of the guitar is good but the keyboard sounded realy awful, just like a Casio toy keyboard pretending to be a piano. Like truely awful sounding, which is when I prompted the AI to try to fix the tone and then it basically just removed it.
The drums were also waaaaay too prominent so I asked it make them a bit more subdued in the mix and it just ended up slowing down everything to the point it just kinda sounded like generic radio alt-rock instead.
But basically once the keyboards were forgotten no amount of prompting could “convince” it to bring them back.
I tried Suno a few months ago out of morbid curiosity and it was waaaay better than this. Actually got something that made my musician friends actually kinda nervous.
If Google can't see the difference between this and useful, moral AI tools then I worry for their path forward.
The music sounds decent, I feel like its missing some things, to be fair Suno still doesn't know what a Puerto Rican guiro is. I assume a lot of these AI platforms will take many iterations.
Things Suno needs to figure out and maybe Google now too, is how to let someone pick a specific voice, and get a rather unique voice, I've heard a few songs in Suno with similar voices to my own songs, and its kind of weird.
I do love making the songs as a hobby, so not a big deal. All in all, AI music is really fun to toy with, especially blending genres together.
One very noticeable difference against Suno is Google Flow Music lets you make Music Videos, which I have yet to test. I wonder if I can use my Suno songs to make music videos for them, not sure I'm vibing with Google's Music AI yet.
Aside: Makes me chuckle a little, since "Flow Music" is a reggaeton catch phrase by Arcangel who would always say "Flow Music" even though it was always called "Flow Factory" he would call it Flow Music.
Edit:
There's some awkward factors Google will need to work out, while the instruments and voices sound nice and clear, the rythm feels weirdly off for some songs, its like the voices are not matching the genre mix, its also missing some nuances I've asked for, I assume it does not know what "wobble bass" means. Suno lets you describe nuanced specific sounds or instruments and uses them how you describe.
I told it to have a dubstep breakdown in the middle, it keeps the artist singing / rapping, which is bizarre, that's not how a breakdown would be...
Suno takes great care to make sure the voice always matches whatever is going on with the beat, including humming the beat / bass / brass / whatever instruments are being played.
Glad Suno is going to have some real competition, I just hope Google doesn't kill Suno with its bigger wallet, would be a shame.
Edit:
Final verdict from me is, it feels like less polished than Suno in terms of music, but more features. Suno lacks music creation which still annoys me they let you make a lyric video that's one single orientation / resolution, you have zero control over it otherwise.
There's a "workspace builder" which you prompt and it builds a web app that lets you create songs and what not, not sure what all the features of it is, but it is interesting as well.
If they get this more on par with Suno, they might for the first time ever take money from me since I left the Android / Google platform many moons ago.