BEST AI-generated EDM EVER
HTML-код
- Опубликовано: 14 окт 2024
- I did AI-generated baroque music, jazz, and rap music before this one, so I figured I'd continue the AI-generated music series with this video! (Yeah it didn't go according to plan, but I decided that I wanna make videos about my failures as well since it's about the journey and not the destination.)
im so exhausted
If you're living in the United States and over 18, go here to see how to vote! / howtovoteineverystate
(I first heard about this RUclips channel because Hank Green linked to it.)
RunwayML: runwayml.com/
The music is either my own (the AI-generated stuff), or "The Lift" by Kevin MacLeod at incompetech.com/
• Kevin MacLeod [Officia...
The hottest album this year "screams of the damned"
Yes
screams from hell
SCP 096: *screams intensify*
The songs sound perfect for a horror game so we got some new sounds for horror games
Lsd bad trip at the club simulator
the song is so pleasing to the ears, it totally doesnt sound like humans suffering from the eternal pit of unending darkness
now we know what concerts will sound like once the AI overlords take over
@@carykh hello there
@@carykh hi
Why Is The Devil In My Room?
Edit: He's Still Here.
2nd Edit: He is still here, send help!
3rd Edit: Now I'm in hell, I guess this is my life now.
it sounds like an SCP that sucks you into the speaker to scream for eternity
bro, when computery gonna drop an album, we’ve been waitin’
Here before this becomes top comment
@@pixelGaming146 good point, hopefully this will become top comment that way the first few comments to that comment is a cluster fu- umm
A cluster comment
SO TRUE I NEED IT
fr :Z
This video isn't even 10 hours old
The title of this video should be. "I tried creating EDM using AI and opened up a demonic portal instead"
ik
Why does this sound like the plot to the DOOM series?
... that's ... That's a UNUS ANNUS TITLE ... Hold On a second
Your profile pic looks like a portal
doom 3 demonic portal not doom
6:57 Thanks, you invented Hell Core.
ever heard of everywhere at the end of time?
It has a name and it's called either dark ambient or noisecore
it genuinely sounds like an oceangrunge base to me ngl
you mean clipping. and death grips?
nah, this is just infernal ichigaya from smt iv
...I didn't know Cary was the _real_ composer of Amnesia: The Dark Descent.
8:32 lip sync failed
Cary was the real composer. Nice.
Cary was the real composer of Amnesia: The Dark Descent.
*Computery
Its just a burning memory...
That first track sure was EDM!
EDM:
E: Portal
D: to
M: Hell
Eternal
Damnation
Music
@@corlinfardal the real gold is always in the replies
@@corlinfardal i agree
@Corlin Fardal lol
Elegant
Death
Machine
New EDM subgenre unlocked: Hellcore
Hell yeah! I’d love to listen to that!
And all it took was a blood sacrifice!
check out hellektro
aka aggrotech
It's just called Noise
@@pyl0nph0togr4phy7 This is _Advanced Noise_
It sounds like the final album in Everywhere at The End of Time.
Ikr
7- Purgatory
Dont remind me of that pls
I wouldnt say so. Its more like Navicon Torture Technologies or Merzbow than it is EatEoT
"I'll be that sound that gets stuck in your ear for months on end"
I already have tinnitus, thank you very much.
Underrated comment
Top 10 RUclips comments
Sure do love the constant unending sound.
Totally couldn't live without it.
First song: *devilish sounds*
Auto-generated subtitles: _A p p l a u s e_
The AI is applauding the other AI's music.
They're communicating
@@The_mario_fan oh no
Oh, it's applause, alright. Just not applause that any human can produce,,,
applause from hell
I actually thought it was the audience cheering at first 😂😅
this music is terrifying, i felt there could be a jumpscare any moment when that was playing
boo
@@ne01nvader GAH
@@mrjellyfish2 did you get spooped?
@@ne01nvader yeah, you can't scare me like that, man
@@mrjellyfish2 no worries bro, I have coke.
Calm down and chill
some people say "demonic screaming" to describe general loud noises
this is one of the few times where it isn't that loud
Turn up the volume song, get a good dose of some of day mmmmmff
it doesnt even sound like the spectrgram displayed along with the portal-to-hell-noise, what happened there?
also if it sounds too noisy try turnin up the contrast on the image before converting to audio
thats true, it looks nothing like the hellish demonic portal, maybe he clicked some weird flag on the image->sound ? not sure
it looks the same as the other images but im pretty sure he played the source file instead of the image->sound version.
(also, you make good music :) )
@@VirtualRiot I think the AI has just no feeling for harmonies whatsoever and is placing horizontal lines in more or less random distances. Also, why are you Virtual Riot?
also i don't see how it doesn't look like the image, if you look at the bottom of the image you can see all the stretched out screams clear as day
why are you EVERYWHERE??
Yeah, this was EDM!
EDM:
E vil
D emon
M usic
There is nothing better than EDM!
@@neonhatesjnj Gee calm down its opinions
@@neonhatesjnj I don’t want an argument to happen 😓
@@neonhatesjnj go away spammer, you're annoying
@@neonhatesjnj tf
"Cary K here"
Nice one.
@@timotheatae thanks, you too
Took me a moment to figure it out. Love the "Cary K* H*" jokes.
me: *hums song*
the aliens hearing their distress call:
I used to think about this when i was 10yo
"Damn what if i make a sound that somehow instantaneously travels to the opposite edge of the galaxy and aliens arrive the next moment"
Absolute bangers. Added to the playlist. The hellish scream really hits, definitely Yeezus inspired right there. I feel like the PC really let their emotions out on this one, and its really what we needed today
Creator: My script is done!
Carykh: *MY SCRIPT*
*OUR
Russia!
Ngl the music at 6:33 sounded closer to a FNaF jumpscare than edm
havent heard that name in years
I wouldn’t call that “music”
Oh my god when he played the first AI song I almost had a heart attack it is so bone-chillingly haunting
LY
The EDM community has been real quiet after this video dropped.
Bruh I think skrillex just quit...
THE WINNER OF THIS YEAR’S EDM AWARDS IS...
*drumroll*
“Tortured Souls Screaming in Agony” by Computery!!!
WOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO
👏😂
2:37 why does no one talk about this 😂
Your mo is a good software too
@@GreatPants LOL it is
@@GreatPantsIt's a bit heavy tho
Perhaps what you could instead, is:
- Generate a bunch of samples (grouped into kicks, snares, leads, pads, effects, etc) [Sources available on Sample Pack sites like Cymatics]
- Generate a bunch of midis (grouped into genres because each genre sounds fundamentally different, like how dubstep is very growl-based and DnB is very drummy) [Sources available just by googling it most likely, there might be more specialised sites, but I can't think of any at the moment]
- smash them together into epico songos
What do you think?
I genuinely love the pieces that the AI made they sound chaotic, discordant and could totally be worked into some sort of industrial trap banger
its like the parts of 100 gecs songs w/o lyrics lmao
Good idea, should make a serum wavetable out of it.
They've got a real Nine Inch Nails vibe to them tbh.
6:30
Cary: Let's listen to the track that AI just made, I'm sure it's gonna be a banger!
AI: ight ima open a portal to hell!
Oh boy, I'm so excited to hear Computerys latest hit, "Unending screams from the darkest pit of black hell"
nice pfp
You can always resample them and make them into ambience.
I miss the old school ai like this lol xD
2:35
"That human-kind has to offer right now, besides your mo-"
Anyone notice that?
yaass, had to double check when I first heard it, and it was hillarious both times 🤣
yes 💀
6:33 Oh no, I played this in my forest and I think it opened a portal to hell and now my woods are haunted by demons
CarryKHeck
daaaang how do i own a forest
@@lmva with money; or a empty plot of land, a shovel and some free time (1 or 2 centuries).
@@lmva Or be a bigfoot like me and inherit it
Aw man
So a little bit of insight: It seems like only the magnitude of the spectrogram was used, but not the phase information which is crucial for reconstructing the audio signal. It might be possible to encode the phase information using the hue of the pixel, which might yield better results.
what phase information?
The silent hill fan in me loves that first version of the ai.
Edit from watching through to the end: oh. Oh no, that was the final version of the ai.
You should make a three hour playlist of stylegan’s best songs!
Hey, unironically, this is awesome. It's pretty ambient noise kind of music and it would fit right into an oppressive atmosphere section of a game or something. Will you upload some results anywhere for general download? : O
This music sounds like how I imagine I sound when I sing, so it's perfect if you ask me.
Wow you’re a great singer
Do a featuring with the AI then
When the "K H" joke includes the letter "K" rather than a word that starts with "K"
*Wait, that's illegal!*
Cary Kary Here Here!
Kary Kay Khere
What happens if we listen to this AI generated song in reverse? I mean, it already sounds it is spawning spirits, maybe in reverse it sounds wholesome?
Wow, Cary, they sound like the best EDM songs ever! Screams of agony from the gaping maw of hell is now my favourite song ever.
Maybe If you try to train the AI models separately for different elements of a music, and then layer them, it might work better.
A separate model for beats, melody, bass etc. In real life, we can't just compose music with all the elements simultaneously. Hopefully, by splitting it down to more fundamental levels, the model can find it easier to train on and actually generate something.
“I’m gonna endlessly blast it into your ear until you go vote.”
Me, who is too young to vote:
Get outta here little kiddo
@@jimmyjohnjoejr
N o
@@jimmyjohnjoejr prob like 16 or 17
@@JesusChrist-lv7uo thank you Henry stickmin
@@jimmyjohnjoejr shut up, kids dont have to get out though
7:13 The start of this one actually sounds like EDM. Recorded on the shittiest mic to ever exist.
Cary; “The song will keep playing until you vote”
Guess I’ll have to go deaf for 2 years
VOTE >:(
@@ThomasBomb45
They’re 16. (If I’m interpreting the comment right)
@@theawesomepanda1lance241 or they're not from the US maybe?
you voted now? its been 2 years
R.I.P Cary's Ubuntu Partition. :(
My original prediction was right!
He did end up doing EDM like I guessed during the rap one
carykh: ok RunwayML make me some music
RunwayML: *demonic hissing*
Thispersondoesnotexist: uhh... exists?
Me: Let’s make ships!
Thisshipdoesnotexist
@@cellohowslife5817 aaa aaa aa aa aa AAAAAAAA
“Cary K Here”
Ohohohohohohohoho I see what you did there
H
@@shekeler *H*
@@ne01nvader H
@@lmva H
Well actually i was actually looking for a way to create demonic screeching.
6:33
Endermen do be making good music tho
I'm no Einstein but think the plural for Enderman is Endermen
I can’t believe Cary would steal, truly an unexpected, horrendous, and catastrophic event 4:00
OUR Code.
@@dennisunderscore *insert an image of a cartoon rabbit standing before a soviet union flag and saying "we"
@6:33 this sounds amazing. It sounds like a hidden track on a Nine Inch Nails album. It's so unsettling. It sounds like a first communication with an extra-terrestrial intelligence.
I didn't know that computers could communicate with the souls of the damned now
I'm an ambient musician and i genuinely unironically love the sound of those AI spectrograms lmao.
Glad to see that ARSS is still being used! I use it to edit out background noises from vocal tracks, and the inevitable occasional click and pop from buffer underruns (that I can't seem to avoid, no matter what settings I use). I do have to band limit everything I run through it specifically to avoid the blurriness that occurs with the lower frequencies, but the clicks and pops are really only most noticeable in the higher frequencies. I can use more conventional editing techniques such as equalization and copy/paste of wave cycles for the lower frequencies. I mix the lower frequencies from the conventionally edited clicky poppy tracks and the higher frequencies that I have edited in paint dot net, and the results are usually kinda acceptable. Many of the results still work with pitch and time manipulation softwares.
That or sometimes I hand draw the lower frequencies. But those results are not usually editable with automatic pitch correction softwares.
kary: "the biggest playlist ever"
Helekjanta's "my music": What's that smol guy out there?
I love your ai generated stuff~ especially music 🎶 I'm glad that your doing more music!
i get why the generated music is odd to people but i honestly really love it
This "music" is the halfway point of the bleeding screams of the dead and EDM.
When you accidentally reinvent Everywhere at the End of Time
Dropping in to say your videos have helped/inspired me to make progress on my first project!!! Thank you @Carykh!
"Most people don't vote, I'm looking at you Millenials and gen z"
My guy most of gen z are children
u-unless that was the joke.
Most zoomers are teens, but early zoomers are adults.
Anyone born before November 4th 2002 will be able to vote in the election this year. The border of where Gen Z begins is shakey but that means anywhere from 4-7 years worth of Gen Z, or anywhere from a little more than 1/4rd a little less than 1/2 of the generation will be voting this year, as a generation is usually about 15 years of people
Yeah, the older end of the zoomers recently graduated college at this point.
Computery hasn't been dealing with the pandemic that well it seems.
6:55, I mean, it sounds like the perfect song for 2020
Yeah, after Covid-19 and more
@@Revilatan Hi Rosettemations
@@masondipperpines5009 Hi?
Cuz its hell XD
Whats funny is that its not static, it actually sound like demonic screaming.
It's much unnerving to see cary tell you to vote while this abomination of a song blasts in your ears, if it was in person i would cry.
Cary has officially taken a break from uploading to be a full time youtube commenter
Yes. But no. Wait actually... yes. No. Yes. But no. Wait yeah.
Oh yes my favorite song "the sound of souls burning in hell"
SuperMario9078 would fit in a horror game
6:27 When biblically accurate angels descend from the heavens
Happy birthday 🎂 cary I love your channels
noise fan reporting. these songs actually unironically slap
The first one sounds like hell is overflowing... Overflowing with SCP-096 clones... And the second one sounds like sirenhead with skitzophrenia...
Alternative title: How to create horror game music using AI!
Everyone talking about the music but I'm wondering about the clip in the beginning where the guy is beat getting with the tennis rackets
That's cary.
Ah yes, my favorite song.
*Ã̷̢̢̢̻̠̯͓͇̯̤̬̰̠̜̥̟̩̗̮͚͍͇̤̦̙͚̝̋͑̃͆̎̎̅̂͑̒̍̏͑̐͝Ă̷̡̨̨̨̡̡̧̛͎̪̞͍̠͈̥͓̰̫͉̖͍̤̖͓̝̞̯̝̩̩̮͙̪̙͙̥̝̱͓̞͈͎̤̣̥͉̭͖̩̮͒̉̓͒̈́̀͛̀̈́̓͌̾̈́̎̐̃͐͐͊̃͆̈́́̾̏̆̿͋̉̾̀̂̿͑̄́͋̆̉͘͝͠͝ͅͅA̴̭̺̦̥̞͔̬̘̺̮̮̱̤̤͇̣̘̽͋̀̐̽͌̍̈́͐͋̊̿͘͘̕͜ͅĄ̸̡͍̩̪̀̈́͆̃̈́̋̈Q̸̨̰̜͉̳̻̱̻̬͇̖̪̣̥͙̲̰̖̙̰͉̳̥̗̙̼̲͎͉͇͖̲̗̻͉̱̀́͆͊̂͆͗̅̃͗̎̋̀͐̿̓̋̆̋̐͆̓̓̄̚͘̚͝ͅͅÛ̴̡̡̧̻̱̞̣̩̠͉̼̪̼̙͚̘͔̘̦̩̭͙̺͓̠̝̠̳̃͆̽̐̔̈̿̐͆͑̏͗́̀̌͂̿̽͝͝Q̸̧̪̬̹̺̠̙̓̿̐̌́͑̈́͊̽̍͘̕͠͝J̵̼̥̟̻̄̿̀́̔͆͊̀̑͐̅̐̔́̀́̇̈̈́͘̕͝Q̶̨̧̢̧̝̟͕͉̥͚̝̗͓̱̼͍̩̰̦̠̹͌͘̕H̶̨̨̡̩̣̘̱͔̳̫̝͓̞͉̫͌̐͑͋̀̂́̕͜͜ͅQ̸̡̨̲̣̻̼͔͐̿͂̏͘Ư̷̘̌̆̂̉̓̓̈͑̾̐̔͒͋̀͗̆͋͋̌̔͝͠͝ͅĄ̸̜̟̗̗̱̮̼̘̤̈́̏̂́̈́̾̿͂͋̓̓̄̔͒͊̔͆͠Ų̸̡̢̧̡̧̧̨̢̨͎̙͉͖̤̣̲̱̻̼̗̩̲̬̞̺̝͎̤̩̱͚͔̦͕̫̹̙͓͓͕̥̠̳̯͔̼͈̰̰̭͙͈̮̤̤͈̺̞͕̞͓̜͇̜̞̰̟͓̺̟͈̰͈̝̟̻̙̩͓͖͙̰̬͍̳̺̝̞̝̭͙̳̤̝̺͔̙̠̫͍̫̠̬̥̟͖̜̫̬̖̼̥̫̥̠͍͉̥̼̩͍͓͎͋̄̂̌̂̈̀͑̽̒̎̏̈̀̓̆̎̈́͑̌̓̈́̃̍̔́̃̒̅̈́͋̾̽̊́̉͆͛̀̑̇̂̕̚͘͘̕͜͜͜͜͝͠͠͝͝͠͝͠ͅͅͅͅĄ̷̧̢̧̨̡̛͉̥͍̬͍̪̮͍͎̠̗͈̗̻̹̞͖̯̝̰͓͖̩̰̝͔̘̭͍̳͉̳͙̥̲͔̗̗̱̹̩̼̩̻̗̝̗͚̜̮͉͔͕̭̘̩͈̝̘̮̜͉̠̋̍̃͂̈́̈́̽͌́͑̇̀̋͑̓̑͋̋̈̊͂͋͌̑̋̓̍̈̊͛̐͑̍͑̑̈́́͌͗̈̇̿͒̓͛͛̈̃͌͛͌͗̂̃̉̉͑͒̉̅͗̃̚̚̕̚͘͜͜͜͜͜͠͠͝͠ͅŲ̸̢̢̨̧̡̧̧̧̛̱̬̝̗̥̱͓̮͚̤͈̤̗͚͓̹͕̟̲͍̼̥͚̳̟͚̤͎͕̗̦̹̲͙̤̪̬͉̻͕̰̜͖̻̯̖͚̮̖̞͕͔̬̦̣̦̰̠͚̝͖̗̥̙̘͇̗̝̤̤̟̻̭̺̤̙̪̟͖͉̼̝̗̳̱͙̭̤̩̠̔̋̔̌̽̊́͛̆̽̓̈̒͒̑̍̋̑̾̀̑́̆̽̐̊̆̌̐̄̓̌͐́̓́̋́̈́̈́͌́̐̋̿̂͒͐̑̈́̍͂̌̔͊̅͌̄̌́̈́̈̌̿͋̒̂̀̅́̓̃̌́͛̉̉̈́̈͊͒͒̇̽̈́̐̋̇̀̅͊̀̾̓͋̌̀̔̉̔͐́́̊͋̋̈́̾̔͑͘̚̚͘͘̚̕̕̕̚̚̚͘͜͜͜͝͝͠͠͠͝͝͠͠͝͠A̷̢̨̢̛̛̛̛̛̛̛̼̰͔͍͓̺̣̹̩͎̲͔̮̲͔͔͚̱̹̘͕̥̘̱͚͉̠͍̟̣͆̈͌̎̀̾̑̓͌͐̐͛̀̂̌̀͊̈́̏̀͑̈̽͗̔̽̎̂̾̍̇͆̂̊̿̓̋̇͐̈́̒́̋̽̉̏̓̀̿̃͒̏̊̿̾̐͋̈̇̍̑̔́͗̋͐̔͐̍͌͌̅͐͗̇̈́͗̂͒̈́̽̆̍̀̓̒̐̄͆̋͂͒̊̈́̿͘͘̚̕̕͜͜͝͝͠͝͝͝Ṵ̷̢̨̧̡̧̡̧̞̤̯̣̖̟̪̬̣̬͖̩͎̫̺̬̖͓̲͔̗͓̞̫̹̩̼̳̳̩̺̳̜̫̹̹̝͈̯̰̪̳͈̺͚͉͇͚͍̤͎̱̝͖̩͔͉͈̫͍͎̘͎͓̪̟͓͙̞̮̗̗̝̻̮̞͇̟̱̠̺̙̱̥̜̤͔͕͍̟͉͇͕̯̫̪͙̻̯͇̤̤͖̹̞͙̗̺̖̫̹͍̝̦̽̓́͂͗̄̇̾̄̌̎̏̀͌̽̄̑͆͋̂̾̆̽̒̕͘̕͘͜͝͝͝ͅͅƠ̵̡̧̡̢̢̢̖͕̯̲̹̬̻̖̥͓̱̮̟̙͓̩͓̟̩̯̜̮̳͙͍̫͈̯͇̙̩̺̹̼̜͖̣͕̩̙̮̝͇͖̺̱͈͙̙͇̬̹̳̹͓̜͙͕̪̋͆͊̀̌̈́͑̋̓̄̓͛̊̀̆̍̊͋̏͐̑̀́̀͌͂̌̏̈́̑̀̃͗͒͛̍̍͋͊̋̐̍̊͆̎͌́͗̉̄̌͆̈́̍̉̍̓̆̆̀̎̔̃̆͊͒̈́̿͂̀̾̏́̎́̉̀̚͘̚̚̕͜͝͝͝͝͠͝͠͝A̴̢̨̡̧̛̠̭̼̟̠͙̝̟̟̪̫͇̭̙̲̹̜̻̭̮̥̠̩̺̼̥̙͕̠̭̲̙̦͉͍̥͇̲̬̳͎̮͎̩̝̠͓̙͓̝̟̹͍͔̰̬̙͔͖̲͕̋̽̉̈̉̀̀͛͐̃̕̕͜ͅƠ̸̧̢̨̧̧̨̡̢̨̡̻̮̟͖̻̖̰̹̗͇̠̮̮̳̝̬̝̜̩̪͙͙̱̫͚͕͈͈͇̼̼̤̲͎̘̩̜̭̦̘̞̻̪͕̟͉͓̖̖͖͇̦͔̩̝̪̖̮͚͉̹͖̮̭̤̘̫̟̭̺̜̩̱̮̟̪̞̖̠͙̯̠̠̝̟̭̙̹̭͈̩̞͕̼͙̰̯̲͍͕͇̩̺̤̫̲͔̱͍͈͙̘͈̖̓̈́́̊̾̽̽̋̓͐̀̿̇̇͗́͌̏͊̾̾͛̍̎͋̊̎̑̌̃̿̿̎͆͊͋̌̔̑̋̍͘̚͜͜͜͠͝͝͠͝͝͠͝ͅȦ̷̢̧̛̛͇͎̱̖̬͍̬̝̭͈̜̊͛́̒̃̈́͌̋́͑̈́̇̈́̓̈̀̀̈̊̒͆͊̆́̊̆͂͐̑͋͗̈̾͗̀̑̇͛̐̆̀͒̃̋̾̃͗̀̐̆̄͒̈́̉̾͒̑͛̀̒̈́́͊̿̀̿͐̂̈́̾̇̔̆̅̍́̀͂̃͛̆̋̑͆̒̏̄͑̔͐̃́̓̃̎̊̀͘͘̕̕͘̚̚͠͝͝͠͠͝͝͠͝H̴̨̧̛̛̛̛̜̦̙̫̗̝͖̳͙̯̯̱͌̇̏̈͂͌̀͂͌̈́͛̾̓͗̋́͆́̈́̐͗̃̾͒̀́̀̀̈́̇̎̒̄̌̍̀̊͊͆͛́̽̒̈́͂͋͒̾̋̎̓́̔̂̿̒͐̓̽͐́̿̾̓͐̃̈́͆̈́̏͌̇͒͘̕̚̕̕͠͝͠͝ͅQ̴̢̨̨̡̢̢̡̢̢̨̢̭̼͓̘͔̲̠̖̫̼̝̻̜̫͉̗̥̤̹͔̭̹̲̣͈̗̣̰̺̹̫̹̝̠̰̱̭͔̣̹̰̰͇͙̱̳̱̻̥̭̻̪̳͚̣̼͍͚͚͎̹̭̯̪̰̳͇͎̘͔͕̣̩̗͈̗̺͈̫̞͚̩͚͎̩̪̝͕̠̜̤̲̣͉͈͕̰͇̼͖̯̰̤͙̟́̅͋͂͋̇̈̒̈́͋͆̀̉̏͂̓̉̎̊̀̑̔̀͒̏͊̊̒̿̇̏̊̽̾̒̋̈́̌́̊͑̐̔͘̚̕̕͜͜͝͝͝͠ͅͅͅQ̵̢̧̡̡̧̢̧̢̨̛̛̛͎͇̲̩̰̲̘͖̼̤̗̦̱̩̗̝͕̲̣̰͙͚̫̹̩̻̫̖͈̠̮͈͖̺̯̭̟͓̜͚̲̳͖̳̘̺͇͔̲̬͚̤̯͓̖̤̯̘̫̠̙̱̱͔̹̻̣͚͈͈̯͚̖͓̯͎̹̖̹̹̲͖͇̩̭̝̼͙͈̯͈̗̲͙͉͍̳̟͕̭̙͈̱͖̙̜̲͍̟͙͙͕̙̟̩̗͍̻̰̮̙̅̏̆̏̈̾́̊̋̎̊̇̇̌̓͑̋̊̀́̉̔͐̃̌̆̅̋͒̔͌̈́̓̀͆̅̇̊̽͌͗̒̆̃͌̉͂́̆̿͛̄̄̉͗̀͂̾̐͊̌͌͐̋̐̿̿̍́̒̆͐̐̅͊̿̏̉̈́͆̿̍̔͑͆̍̑̂̈͌̓̄̃̏̍̾̅̈̈̚̕͘̚̕̕̕͘̚̚͜͠͝͝͝͝ͅͅͅQ̵̡̡̢͚̥͙͇̦̗̖̬̪̣̩̥̱͓̠̟̗̞̼̺̫͖̟͚̝͎͖̬̘̜̗̯͎̝̼̤͓͓̤̤̟̮̺̣̩̻̤̠̜̜̭̬̰̞̗̘̜̯̘̰̞͍̠̝̼̻̥̯̤̤̝͙̹͈͚̞͉̦̊͋̓̔̋̀̅̽̈́̈́̓̎̀̋͋̀̋̍̿̆͒͋̔͒̏͠͝͝͠ͅͅQ̷̢̢̧̢̘̬̫͓̞̜̣̪̫̼͍̫͍͉̟̫̲̖̯̜̳̞̝̦͙̹̩͓͚̼̲̟̥͓̻͇̩̺̳͚̠̰̲̮̩̬͈̼̭͎̊̓͑̍͑̂̒̃̔̔̿̀̓̅́̓̽̄͒̒̽͑̓͂̈̇͑̇̈́̃̈̈́̈̍̈́̃̓̄́̉̊̀͂́̀͌̾͆̈͂͛͒́̓̄̐̃̿̑̾͐͐̉͜͜͜͜͝͝͝ͅ*
Wow, this is very witty! Great idea and great job persisting with all this work. It's a pity it didn't go well at the end, but that's not the most important thing. Thanks for the video!
ok boomer
It's a shame that it is now used to generate fake "art" to replace real artists.
agree
4:57
The first thing that comes in my mind: *B A G L E B R A I N*
It literally sounds like horror music, that a horror game could use as ambient sounds, I love that so much
The spectrogram comes from the wavelet transform of the music, which actually uses COMPLEX VALUES.
The biggest problem with your method is that it doesn't include the phase data of the frequency components, which contains the majority of the timbre, instead it only contains the magnitude.
TRY THIS AGAIN but set the HUE of the spectogram to the PHASE. You can extract the phase using the arg() function. Python definitely supports complex values.
After generating it, you can extract the phase using the hue of the generated spectrogram, and magnitude with brightness.
Oh boy, an upload from Cary Kaiming Huang from hit channel carykh!
everyone: wtf
experimental music lovers: dope
Me singing along to my favourite song: nether portal noises
9:28 I'd much rather have people uneducated on politics not vote. At least we have the electoral college to somewhat adjust for it.
Well to be fair it would be better if people got educated and then voted. But it is frustrating that the first half gets left out a lot.
your animating noice
and nice job in general wow this is genuis
youve gained 1 sub
5:54 mp4 is video. Mp3 is sound. The screenshot was taken in audacity which cannot read mp4 files. Sorry if i'm wrong.
but it will only show the audio
@@HairyBowlLithuania this was 3 years ago i get the joke now ♥
*accidentally glimpses into mathematical dimensions of hell
There was no Kh joke so I’ll make one
“Hi guys, Cary Knockout Human here”
There was a joke he said cary k instead of cary kh which really isn't a joke-
@@orcacultist4526 no it is a joke, he said "Cary K Here" "Cary K H"
this song is a bop to my inner demons
Nice video Carykh!!
Fascinating work! That song could probably be used as a sample for a human musician's song so I think it's fine that we're still in an 'AI-assisted' art era. The training wheels are not ready to come off, haha!
I know that this is out of topic, but can you make a video about using gentle( a robust yet ... ). I am really having trouble installing it.
yeah plz
@@imbleshagadamba9575 I am suffering to download and install it!!!!!
Yo, where can I get that... uhm... "music"
I wanna play this tonight.
OMG! tysm!!!!!
Great new additions to my study playlist ☺️😊!
I’m looking 👀 forward 👉 to 2️⃣ more fire 🔥🔥😤💯💯😩 and relaxing 😌 musics to listen 🎧 to while working 🥶🥵😳✊✊😤😤😔💯🔥💯😔😤😔💯😔💯💯😔💯
That first track was the embodiment of the sound the subway makes when it's stopping
Dude. Props for encouraging people to vote!
It won't work because the AI has no concept of music theory, and most importantly the EDM dataset is too diverse with too many different pitch keys and musical scales; you'd need to program in very specific rules, or have a controlled dataset to get a result resembling music.
But that's exactly what a good ai learns. They don't need music theory. Humans don't need music theory: just listen to 1000 songs and you can figure out your own song. That's just what ai does
@@SandwichMitGurke Good ai still needs proper guidelines in how it's programed, and some rules to steer its learning. What I'm saying is the video's methodology is flawed IF the intent was to make traditionally pleasing music. If not, it's still a good exercise on how good or bad unguided ai is.
And yes, works are derivative and you can make music from examples, but the effectiveness is limited if there's no capability to understand music's mathematical principles relative to our psychology, like what the frequency relationships are that makes something sound good to humans IE. Music theory.
@@GENKI_INU
Most AI that manages to recreate music don't know music theory either, and some of it sounds really good. Future versions will be even better, all without music theory knowedge.
@@cortster12 Yes, with proper actual controlled samples you feed it from someone who knows what they're doing with music theory, and also a positive and negative reinforcement system to help point the AI in the right direction over many generations. That's the only way these experiments will be productive. I can tell you the psychological (psychoacoustic) aspect of music is highly technical if you want good results.
0:01 haha I see what u did there
0:32 "GPU overlord"
Ryzen 6000 is behind you
An interesting experiment, but yeah, gradient optimizations lead to a lot of smearing/blurring in the results which works better for more continuous type signals and less so for discrete ones. Would be interested to see how this works/compares using midi files(discrete steps) as input/output instead of full sonic spectrum and varying time-window sampling intervals to capture time-interval repeating patterns/features.
The best thing about it is that you create totally unique sound effects for free! You can do whatever you want with it :)