idrc, I don't like any songs with foid vocals, I don't care if they're good or not
I love technology, can't wait for the day when robo girlfriends would be a thing, I'll marry one for sure, even if I have to sell my entire generational wealth.
B
View: https://www.youtube.com/watch?v=0gAMeasM0Nc
Tbf though, trance music is the probably the naturally easiest music for a generative program to make.
Yes i posted about suno beforeAI music surprised me, it has developed to be quite decent already. It used to have trouble simulatng vocals, but it's gettng much better at it.Anyone else tried it out?
You can try it out on this site:
You can select to either make a 'Custom' song (with lyrics you enter), or to generate lyrics with ChatGPT based on a prompt. I prefer to use custom lyrics, but it's your choice. Then fill that in, specify a style for the music, and it will generate two songs for you.
It has a few very basic censors on the lyrics, but so far generally nothing serious. It checks for certain words, but you can generally get away with homophones, etc. This will probably get worse as (((certain people))) demand more censorship, which is why things like ChatGPT became almost useless for art.
Here's some examples I made, with custom lyrics:
Closer by @technicalhouse596 | Suno
sad stripped back alt rock. based on joy division song. Listen and make your own with Suno.app.suno.ai
For each prompt, it makes two versions. You can pick the one you prefer or make more. You can also continue them, if the two minute default length isn't enough.
The AI does have flaws. Sometimes the AI gets a bit confused, eg. it garbles the first line of the next one. But the rest of it is fine. I think most songs I've tried with the AI haven't had serious issues with parsing the lines, although when given some subgenres (eg. 'nu metal') it will stick to a very narrow interpretation of the subgenre.
Blakkkk Ocean Spiral by @technicalhouse596 | Suno
a soulful, angry acoustic track, vocals based on william gold song. Listen and make your own with Suno.
(^ a touching song. 'You say I'm a fakecel, You say I'm a gray, You say I'm not worthy of you, Stacy...' Also a more metal version here, it's a bit inexact but the beginning is based: https://app.suno.ai/song/cfaca653-2116-4d14-b2ea-9cdb35d92f48 )
View attachment 1106926
(if Rise Against had balls ig)
Anyone else tried this stuff out? It mogs incelcore bands IMO.
Post your own songs if you want.
made a few songs last weekend:
^ Inspired by Twilight uWu
islampop is underrated. @Darth_Aurelius thoughts?
^ Inspired by Twilight uWu
islampop is underrated. @Darth_Aurelius thoughts?
Last edited:
'St. Patrick's Day,' an inspiring song celebrating the spirit of the oirish or whatever:
tbh fuck the oirish.
was asked about female vocals and had barely tried them, so decided to test some ideas
barely any females with decent voices smh, and those aren't obvious options. but the like one or two options can do fine, ig?
in other news related to marginalized minorities, the AI has problem comprehending that white rappers like skitz kraven exist and keeps trying to turn them black. plus the censors were bitching constantly about the lyrics. this is like rosa parks smh
in other news related to marginalized minorities, the AI has problem comprehending that white rappers like skitz kraven exist and keeps trying to turn them black. plus the censors were bitching constantly about the lyrics. this is like rosa parks smh
Last edited:
non-meme AI threads won't get much engagement here until the AI is fuckable, and that's on the whole a damn shame
based tbh
asakura is the margot robbie expert, I bet he'd be able to help create this opus jfl
in fact, he's already made a good start IIRC
https://incels.is/threads/i-created-song-about-margot-robbie.599788/
Yeah but i can only make a FEW songs a day, Suno is really jewish,
13 dollars a week? Maaan GTFO outta here, I need new tools!
13 dollars a week? Maaan GTFO outta here, I need new tools!
I did not see this one. It's good. He tagged me, but it was during my requested self ban.
I thought you were going to link to this:
yeah, agreed on them being jewish af there
atm can get a decent amount of chances by altmaxxing, plus they gave 500 free credits per acc at some point last month. it's possible to get around the censors, usually, but they can be a bitch for certain songs.
on the upside, the amount of copyright shenanigans possible with it are hilarious, they pretty much just let the AI scan through all kinds of shit. there is a basic censor to protect against it, but still possible. though I tend to focus on artists with enough quality to be decent here
omg based song, ty for posting
I did not see this one. It's good. He tagged me, but it was during my requested self ban.
I thought you were going to link to this:
View attachment 1164724
rage against the margot
^ last song was inspired by my struggle (14/88 btw) at making a sludge metal song that didn't degrade into random pseudo-screaming. I sort of gave up on making that song, and screwed around with the vocal lines till I found sth that sounded right for this:
OG lyrics (more based):
Last edited:
song inspired by sheo's odd fling with the weird mid druggie chick:
I ship them, so this is about how idk Sheo will eventually become a god-tier druggie and druggiemog her, then they'll get together or sth like that. it's written in the stars
I have seen the future.
also, I've been experimenting w/ switching up to other foid voices + genres. found one or two that kinda work, if a bit trippy. it's a bit annoying, tho, most foids are dismal. why can't all foids sing? awful.
I ship them, so this is about how idk Sheo will eventually become a god-tier druggie and druggiemog her, then they'll get together or sth like that. it's written in the stars
I have seen the future.
also, I've been experimenting w/ switching up to other foid voices + genres. found one or two that kinda work, if a bit trippy. it's a bit annoying, tho, most foids are dismal. why can't all foids sing? awful.
Last edited:
I like listening to sound-track music, like epic music playlists, I wonder if it can do that
'epic music' as in the kind of stuff that gets spammed in YT playlists and thrown relentlessly into rando low-effort films, video games, etc.?
I'd imagine it could do alright with that, but not really familiar with the genre. I did briefly know one guy who was a fan of it and had long playlists, tho, but I only found the softer, synth/piano ones remotely tolerable. here's a couple basic tries tho:
tbh in terms of soundtrack music, stuff like ecco the dolphin was p. kino
Last edited:
(sort of feel like a hipster for using the word 'mise-en-scène' in a song, ups. brb going to kms (in
Last edited:
ngl this was fun af. also kinda cool how the AI starts to play along in the second verse lmao, it sounds so excited when it describes the 'blacked: george floyd edition' film. based robot
somehow it turned ben haenow into some kind of craven villainous country singer thing.
somehow it turned ben haenow into some kind of craven villainous country singer thing.
fwiw they recently added the option of using version 3.5 on suno, which allows you to make songs up to 4 minutes and continuations of 2+ mins.
3.5's songs are more streamlined and have a slightly better sound quality, but I still prefer 3.0 atp. if you're new to it, it's probably much easier to use 3.5, though
not entirely sure when it was added, since I haven't used it for a couple days, but here's a few songs I made with 3.5 today. no continuations used, so it's pretty much what the AI came up with
if your song's lyrics don't last up to the 3 minute mark, then it will often just repeat a form of the last verse section and chorus at the end. that can be lopped off, but it's left on here as an example:
3.5's songs are more streamlined and have a slightly better sound quality, but I still prefer 3.0 atp. if you're new to it, it's probably much easier to use 3.5, though
not entirely sure when it was added, since I haven't used it for a couple days, but here's a few songs I made with 3.5 today. no continuations used, so it's pretty much what the AI came up with
if your song's lyrics don't last up to the 3 minute mark, then it will often just repeat a form of the last verse section and chorus at the end. that can be lopped off, but it's left on here as an example:
anyway tried 3.5 a bit now, and it can give some decent results
('call me the phantom of the brothel')
my impression is that it's more streamlined and polished, but 3.0 is slightly better when you want an organic, spacious sound. here's some old songs from around mid-may on 3.0, never posted them but they're quite nice:
(^actually, think I shared the above one in a discord DM from mid-april. I wrote it quickly after someone mentioned that pufferfish make underwater crop circles.)
('call me the phantom of the brothel')
my impression is that it's more streamlined and polished, but 3.0 is slightly better when you want an organic, spacious sound. here's some old songs from around mid-may on 3.0, never posted them but they're quite nice:
(^actually, think I shared the above one in a discord DM from mid-april. I wrote it quickly after someone mentioned that pufferfish make underwater crop circles.)
Last edited:
also, linking a good spanish song thread by @Moroccancel2- :
incels.is
and here's a few other decent robotsong threads, for reference:
incels.is
(a song by uggo mungo abt vivian)
incels.is
(diglet's tracks)
incels.is
(asakura made a song about his arch-nemesis margot robbie lmaoo)
Blackpill song - AI generated (lyrics by Spanish brocel)
View: https://voca.ro/1dJVEo0mfA3F Spanish lyrics: Viste que tú eras la oferta, de aquel mercado exigente; Entendiste acaso triste, tu valor inexistente; Sufriste como Dios te hizo, en un dolor penitente; Comprendiste tras esfuerzos, que todo es aparente; Sentiste no valer nada, ante una...
incels.is
and here's a few other decent robotsong threads, for reference:
Vivian, she has a PHD, Vivian, she studies you and me (rate AI song)
(AI didn't generate lyrics but generated the music) [verse 1] Oh, Vivian, vivian face like an alopecia inflicted simian, upon a print out of your busted face, to my shame I must kooooom, registration leaves you disgraced tech illiteracy is your doom [chorus] VIVIAN SHE HAS A PHD VIVIAN SHE...
incels.is
Banger AI tracks I generated
incels.is
I created song about Margot Robbie
https://suno.com/song/18d15f5a-ae9d-451b-8eb8-8c06f8e5f413 @KillNiggers @wereq @NeverEvenBegan @anandkonda
incels.is
Last edited:
The second one is closer to what I meant. It could be a horror soundtrack. The others are too quiet.
Does the new version work better for epic music?
yeah it reminds me a bit of the kevin moore soundtrack for the turkish film uyku
ok GrAY
don't know, I haven't tried it. 4 mins for filler soundtrack music could get a bit stale, but you can try it if you want
3.5 has its flaws, but it can give decent renditions of ice nine kills-like stuff which 3.0 had issues with:
ice nine kills make good horror-themed metalcore-adjacent music. most of their output is kind of campy filler shout-outs to random horror films but they have some solid songs like welcome to horrorwood, savages and um stabbing in the dark.
the AI generally seems to associate them with the campy horror side, but that can be fun too.
other than that, the time limit is now 4, but I'm kind of used to keeping within 2-3 mins. most genres which suno favours tend towards shorter song lengths, anyway, since it tends to be v. vocal-focused and the easiest way to get a longer song is just a word-heavy genre like rap. so ig you could just write rap songs
Yeah, you've seen them about, the rap singers? You've seen the rap singers, on the - on the Top of the Pops they have them, don't they? They used to come on there, didn't they? The rappers, the rap singers come on the Top Of The Pops.
You've seen them on there, there'd be a young woman singing, and then one of the rappers would come on and they'd talk, and do a little dance. You've seen them, the rappers, the rap singers on the Top Of The Pops now. They're on the on a lot of the adverts now. They might have a rap singer on one of the adverts.
It might be a sausage, or some wool or something, and there'll be a rap about it, about how good the sausage is.
You've seen them on the... You've seen them on the Top Of The Pops and on the adverts. You've seen the rap singers, you've seen them, yeah, you know what I'm talking about, the rap singers, and on the films they have them now. You might see a film in the pictures and James Bond or someone, he might go in one of these nightclubs and there'll be a rapper in there doing a rap. You've seen them, yeah, on the Top of the Pops and in the, in the adverts. In the films, they might have them.
And, um, in the shopping centre you see them as well, don't you, the rappers? They run along the, um, you know, there's like a hand rail — a banister along the steps there.
now yeah maybe writing only long rap songs isn't my thing
instead I just thought of quickly practicing with something that could drag on to ~3k words and is within the genres which suno favours (and isn't a rambling monologue by stewart lee), and it's like, well a teenfoid's diary entries could drag on forever and modern pop is often interminable diaristic foid shit. so yeah.
(when using a somewhat generic foid voice in style prompt, the AI seemed to draw on previous songs it had generated as well. which worked fine here + was appropriate, but funnily 3.5 has problems producing the same thing when prompted to.)
this song was informed by a deep literary engagement with the diary of anne frank:
3.5 does have problems with a few things that 3.0 is good at, as I said it's a bit worse for more spacious and raw artists. plus 3.5 sometimes gives a different take on a vocalist, which is often true to their most recent works but not always what works best. so I still use 3.0 a bit, it produces better results for some songs.
ice nine kills make good horror-themed metalcore-adjacent music. most of their output is kind of campy filler shout-outs to random horror films but they have some solid songs like welcome to horrorwood, savages and um stabbing in the dark.
the AI generally seems to associate them with the campy horror side, but that can be fun too.
other than that, the time limit is now 4, but I'm kind of used to keeping within 2-3 mins. most genres which suno favours tend towards shorter song lengths, anyway, since it tends to be v. vocal-focused and the easiest way to get a longer song is just a word-heavy genre like rap. so ig you could just write rap songs
Yeah, you've seen them about, the rap singers? You've seen the rap singers, on the - on the Top of the Pops they have them, don't they? They used to come on there, didn't they? The rappers, the rap singers come on the Top Of The Pops.
You've seen them on there, there'd be a young woman singing, and then one of the rappers would come on and they'd talk, and do a little dance. You've seen them, the rappers, the rap singers on the Top Of The Pops now. They're on the on a lot of the adverts now. They might have a rap singer on one of the adverts.
It might be a sausage, or some wool or something, and there'll be a rap about it, about how good the sausage is.
You've seen them on the... You've seen them on the Top Of The Pops and on the adverts. You've seen the rap singers, you've seen them, yeah, you know what I'm talking about, the rap singers, and on the films they have them now. You might see a film in the pictures and James Bond or someone, he might go in one of these nightclubs and there'll be a rapper in there doing a rap. You've seen them, yeah, on the Top of the Pops and in the, in the adverts. In the films, they might have them.
And, um, in the shopping centre you see them as well, don't you, the rappers? They run along the, um, you know, there's like a hand rail — a banister along the steps there.
now yeah maybe writing only long rap songs isn't my thing
instead I just thought of quickly practicing with something that could drag on to ~3k words and is within the genres which suno favours (and isn't a rambling monologue by stewart lee), and it's like, well a teenfoid's diary entries could drag on forever and modern pop is often interminable diaristic foid shit. so yeah.
(when using a somewhat generic foid voice in style prompt, the AI seemed to draw on previous songs it had generated as well. which worked fine here + was appropriate, but funnily 3.5 has problems producing the same thing when prompted to.)
this song was informed by a deep literary engagement with the diary of anne frank:
3.5 does have problems with a few things that 3.0 is good at, as I said it's a bit worse for more spacious and raw artists. plus 3.5 sometimes gives a different take on a vocalist, which is often true to their most recent works but not always what works best. so I still use 3.0 a bit, it produces better results for some songs.
Last edited:
'Whore City'
^ DRAIN THE SLAG SWAMP MBGA
had to listen to some drunk lesbian nigger whining abt how her bff was 'stolen' by a guy, and how she hates having to act like a guy to impress other girls or sth. may have influenced the next song idk
****
^ DRAIN THE SLAG SWAMP MBGA
had to listen to some drunk lesbian nigger whining abt how her bff was 'stolen' by a guy, and how she hates having to act like a guy to impress other girls or sth. may have influenced the next song idk
****
Last edited:
sort of old song (mid-may) from 3.0, didn't expect a fred durst impression to be this aggressive lmao:
sadly quality on 3.0 is often down since the new version was released, which can be a bit of a nuisance
but 3.5 is capable of picking up on a few slightly complex prompts which 3.0 could get thrown by
Last edited:
Some news for sunocels:
www.rollingstone.com
Suno was sued for copyright infringement by major labels, along with alternative AI site Udio. It's not surprising, the music business is renowned for being censorious about copyright, and sites like Suno didn't really hide that they rip off large amounts of copyrighted material
All the same, it would be a pity if this resulted in further censorship and limitations to music AI in the future, which will probably gimp them in many unforeseeable ways. Censoring AI generally reduces their basuc functionality, not just their ability to perform the specific prohibited behaviours
will miss the era where it could just rip off random pop singers without blinking a roboteye. sadly 3.5 is already worse at it than 3.0 was
although I never really saw much point in plagiarizing TS, find her voice grating and idk how adults listen to this creature who is blatantly manufactured in a lab to lead singalongs at 5 y/o kids' birthday parties
speaking of obvious ripoffs, wrote this song a while ago in early may or sth during a discord chat. it was a bit screwy at the start + end, but I like how it eventually turned into some kind of simple plan incel anthem:
Other than that, they added sth for subbed users that lets them use uploaded audio for their music. I haven't tried this yet, and it's apparently quite erratic. Tbh idc that much abt it, AI music is robot-generated and selfish humans should let robots do it
anyway some other recent-ish songs:
AI's Most Ambitious Music Generators Accused of 'Massive' Infringement In New Lawsuit
The Recording Industry of America sued the AI music generators Suno and Udio on allegations that they infringed millions of songs for their tech.
www.rollingstone.com
Suno was sued for copyright infringement by major labels, along with alternative AI site Udio. It's not surprising, the music business is renowned for being censorious about copyright, and sites like Suno didn't really hide that they rip off large amounts of copyrighted material
All the same, it would be a pity if this resulted in further censorship and limitations to music AI in the future, which will probably gimp them in many unforeseeable ways. Censoring AI generally reduces their basuc functionality, not just their ability to perform the specific prohibited behaviours
That seems like a fairly clear case, it is v. obvious that Suno is trained on all sorts of copyrighted music and can emulate that with the right prompts. while AI legislation is still not set in stone, this will likely result in AI getting diluted in the future unless it's open-sourced
will miss the era where it could just rip off random pop singers without blinking a roboteye. sadly 3.5 is already worse at it than 3.0 was
although I never really saw much point in plagiarizing TS, find her voice grating and idk how adults listen to this creature who is blatantly manufactured in a lab to lead singalongs at 5 y/o kids' birthday parties
speaking of obvious ripoffs, wrote this song a while ago in early may or sth during a discord chat. it was a bit screwy at the start + end, but I like how it eventually turned into some kind of simple plan incel anthem:
Other than that, they added sth for subbed users that lets them use uploaded audio for their music. I haven't tried this yet, and it's apparently quite erratic. Tbh idc that much abt it, AI music is robot-generated and selfish humans should let robots do it
anyway some other recent-ish songs:
Surprised @AsakuraHao & @ChudMasterofBation haven't touched this thread.
Good call, Asakuro has some bangers
I think they prefer to make separate threads for each song, tho
I've collected a few thread links here, maybe at some point I'll try to index most of the AI song threads since it's easy to lose track of them. and some are pretty good
based
!! NEWSPOST !!
(jfl yellow looks so weird on this site)
so they recently made the 'use uploaded audio' feature available for all users
this means that you can now upload audio of your own, and the AI will be able to create an extension of this which is based on it
not sure when they did this, I just found out yday but I hadn't been using the AI for a bit
this feature could probably be used for based copyright infringement purposes, but I haven't tried that yet
I just made a quick 30-sec voice recording to test it out, and recycled some lyrics I had already used. it tends to use its own vocal archetypes rather than your input, but it does use your input as a reference point and adapts the voice to it. got some decent results from it
female:
(used minimal prompts, so that it stuck to the uploaded audio a bit rather than just going off and doing its own thing. just took some random lyrics I had used recently for this, since I was mainly testing the new feature out)
male
(partial, since I've already posted a version of this song in the thread, and tbh it's mostly further from my voice than the foid song):
**
found foid voices easier here, bc they're less obtrusive and more adaptable.
when using male vocals, it tends to default to a generic archetype with only occasional references to the audio input. for example:
the best way to use it I've found it just making hybrid songs (hybrid theory is law), where you specify a voice etc. but your voice is used as a reference point to enhance that. it's a bit like a usual AI song, but it does take your voice as a reference point and can sound fairly different from just the style prompt itself. here's an example:
had that song lying around, but wasn't able to get a version which I liked. this helped quite a bit, so it might be useful as a way to give the AI some help when it would have issues
anyway, have been going over a few old songs just to experiment with hybrid prompts, and have a few decent results by using uploaded vocals to support a song with an existent vocal style-prompt:
(jfl yellow looks so weird on this site)
so they recently made the 'use uploaded audio' feature available for all users
this means that you can now upload audio of your own, and the AI will be able to create an extension of this which is based on it
not sure when they did this, I just found out yday but I hadn't been using the AI for a bit
this feature could probably be used for based copyright infringement purposes, but I haven't tried that yet
I just made a quick 30-sec voice recording to test it out, and recycled some lyrics I had already used. it tends to use its own vocal archetypes rather than your input, but it does use your input as a reference point and adapts the voice to it. got some decent results from it
female:
(used minimal prompts, so that it stuck to the uploaded audio a bit rather than just going off and doing its own thing. just took some random lyrics I had used recently for this, since I was mainly testing the new feature out)
male
(partial, since I've already posted a version of this song in the thread, and tbh it's mostly further from my voice than the foid song):
**
found foid voices easier here, bc they're less obtrusive and more adaptable.
when using male vocals, it tends to default to a generic archetype with only occasional references to the audio input. for example:
the best way to use it I've found it just making hybrid songs (hybrid theory is law), where you specify a voice etc. but your voice is used as a reference point to enhance that. it's a bit like a usual AI song, but it does take your voice as a reference point and can sound fairly different from just the style prompt itself. here's an example:
had that song lying around, but wasn't able to get a version which I liked. this helped quite a bit, so it might be useful as a way to give the AI some help when it would have issues
anyway, have been going over a few old songs just to experiment with hybrid prompts, and have a few decent results by using uploaded vocals to support a song with an existent vocal style-prompt:
Ye
Similar threads
Users who are viewing this thread
- SoycuckGodOfReddit
- BornToLose
