o-oh… y-you don’t… c-care…? (╥﹏╥) i-it’s okay, i guess… i just… i w-wanted to share something i worked really hard on… b-but… i-it’s fine… i’ll just… sit here… in the d-dark… w-with my spectral w-watermarks… a-alone… (´・ω・｀) d-don’t worry about me… i’ll be okay… maybe… (っ- ‸ - ς)
Bro cares so little that he took time out of his day to look away from his own pitifully unpopular channel and lack of subscribers to tell us about it.
No, it's still very different. In mono the waves cancel each other out. You can hear a little bit of weirdness in stereo, but that could be part of the song.
You should hide the entire script of your "fallout 3 is better than you remember is bad actually" video in your next video. its like a free video inside your video it's free real estate.
@@idiotiniumI was thinking of using morse code in my first album. Never seen anyone do that before. I don't know if I'll use this method you just showed us but it looks interesting!
@@lilyluhtwizzyi think ive heard only 1 other artist use morse code in their albums in some way if i remember correctly so like if you do it there will be atleast 2 artists that have used morse code so like do it it would be cool
When I was a kid in the 70's and 80's, "the church" was freaked out about subliminal satanic messages in rock-and-roll music. I played so much Led Zep, Queen and Beatles backward and learned how to "talk backward" and make my own. I had a lot of fun inserting backward-masked messages into "church" music. Now we are in the digital age, and your video has inspired me to take the trolling to the next level!!!!
It's not as trippy as it sounds, 3blue1brown has a great video about it. Basically it just means that waves aren't instantaneous. In order to contain very specific frequencies, a signal needs to have a certain length. The resolution of the frequencies corresponds to the y resolution on the spectrogram, and the length of the samples corresponds to the x resolution. So the uncertainty principle says that higher y resolution comes at the cost of lower x resolution
As soon as OpenAI adds music generation to their model Sam Altman will be all over the podcast circuit crying about how music watermarks poison the well.
Years ago ( like nearly 20 ), me and a friend of mine were looking to find a way to copyright music with a similar technique. Objectively, the imprint was inaudible, but would be very easy to see when looked at with the method you show. Ultimately we never took it anywhere. Two reasons for that. One was that Sony had a similar technology for their Play Station systems and DVD's. Second reason is that we didn't believe CD's would last much longer, and that most audio would go to digital format anyway, meaning all the metadata would be a part of it. Cool way to make visual sound though!
ya but its not that great, using an external plugin processes the sound in the frequency domain instead of actually just changing the envelope of the partials
it is quite insane to think that artist could do this back in the early 2000's i wonder how many hoops they had to jump through to achieve the same thing compared to today
thats what i dont get about clubs, why is it mono? i'd understand if they setup the sub to be mono out of a single subwoofer, but the mid and high end being mono too doesnt really make sense to me and the only stages that i've worked on actually had a stereo sub too honestly
@@idiotinium Much easier to adjust volumes etc if the tracks are on a per speaker basis and mono instead of full on stereo or surround on a single track based on the club room.
The subway surfers at the end really helped with keeping my attention on the video!!! Such a lifesaver, wouldn't have made it to the end without it. Jokes aside, this is really cool :O
i love that in order to draw the watermark you actually have to make a sound, it feels like using some sci-fi laser machine in an anime to stamp the watermark on the audio track.
First off, this is exceptional and highly-appreciated content on a subject that's needed now, more than ever. Let me also say that I've been making and producing music (as a hobbyist) for over 2 decades at this point. I started out on FL Studio, and have used many other DAWs, before finally settling on mainly using Ableton. From a theoretical standpoint, I fully comprehend everything you've laid out in your thoughtful explanation. However, I have to imagine there's a less complicated way to attain the results you're showing here, right? The inputs are static: a 2-bit color watermark image of a specific height, and one audio file to watermark. How is it that the process would require downloading and installing and learning and using so many tools? I have so many things to focus on now, I shudder at the opportunity cost of setup and learning all of them. Do I _really_ need Harmor, a spectral compressor, spectral clipping, a stereo shaper, phase inversion, etc.? Ideally (especially with how many tracks I work on), I want to click like three things per project and be done.
harmor and AudioPaint are the only tools i know that can create a spectral image spectral compressor is the way i know of that can compress spectrally like that which is free the spectral gate from spectral clipping certainly helps threshold your image's signal (spectral clipper is free) its baffling to me too that spectral tools are so obscure even tho theyre incredibly simple
This is actually stupidly cool and useful. I imagine you could do this in a way that it also becomes a "signature sound" in tracks as well. Would've saved Queen some issues with Vanilla Ice lmao "Your Honour, if it would please the court, may we switch the tracks to mono audio and review the spectrograph?"
So I could do this... Or make a "signature" sound with that "spectral stamp" and use it in every song. Or just use my producer tag in every song. Or a mixture of the three. Thanks for making this, I never imagined the sheer concept of watermarking could trigger some inspiration in me.
I barely understood what's happening on technical level but this looks like what you said. Clever way to sign/watermark your sounds and music. Inaudible to me after fully polished. Now draw her making garlic bread
Hey since you're doing the 3 thumbnail test thing, I clicked on the black thumbnail but it was the weird one with the face that caught my eye and made me add to watch later, my vote to the weird face one👍
I clicked on this video from the frame my patched app showed me - from 2:23 So here's a vote from me that no thumbnails are the best thumbnails cause I've never seen your channel before this vid
it so insane that everytime im like messing around trying to learn how to do things like this or other weird experimental stuff and then someone uploads a video soon after
you can change this to cancel out side instead, so that its mono compatible and only shows up after inverting one channel and then collapsing to mono to do this, swap mid and side before you send it into the chain, then swap back after (this should work, if not then idk lol)
This is a very neat idea. I’ve thought a lot about how you can watermark demos and instrumentals and stuff but this idea never crossed my mind. Sick tut 💪🏼
I'm surprised at how much of this I actually did correctly when I did this. There's some things here and there that I could've done better but man was I close to getting it perfect Thank you for explaining it
Yup. And a good mix will be manufactured in such a way as to collapse to mono without serious side effects. I have essentially monaural hearing, so I make little combiners that feed both the left and right channels into each ear cup in my phones.
hmmm, isn't that the opposite? Smaller rooms are difficult to emulate stereo because you need to actually separate the L and R so much to have proper stereo? Don't bigger stages nowadays run proper stereo? Even decent clubs run their main system in stereo, sending mono signal only to their speakers outside the main floor (for example bar speakers)?
From what I've heard from different sound techs this is actually a myth. Properly set systems allow for impressive levels of stereo on big concerts and festivals
This is one of those irritating tutorials where you move far too fast and thus despite rewatching and refollowing, mine still doesn't work, and I find another channel that doesn't clear his throat every 3 seconds. Thanks!
That’s quite unexpected and fascinating. Great vid. I will have to think of a way to use that base idea in a less destructive way (the issue cited by many commenters about large sound systems being in mono is definitely something to consider).
@ oh, that’s a great idea. A sort of double-hiding mechanism. The thing I wonder however is how much that scrambling of phases blurs the spatial definition in the stereo mix. The exact opposite of my previous concern, so to speak ! 😁 Sorry, people are always complaining and wanting the opposite of what they have ! 🥲 This all raises a million of questions, like how it behaves if you apply the trick to only one instrument in the mix, or of other ways to embed hidden information in the phase. I can see this kind of processes be used in steganography, for instance (like is already the case in some image processing that remain invisible to the human eye). Fascinating subject.
its not really scrambling any phases, its moreso just separating your signal into mid and size, and then you're just punching holes in the frequency spectrum. either making some regions entirely mono or entirely side (depending on what you're cutting out) if you only apply it to one instrument then that instrument will be the only track that contains the watermark, everything else will just go over it, like if you painted a shape on a wall and then painted more shapes over it
@ I understand all this, I am just wondering how deep one may go before the hidden information get totally blurred. I guess this is important for instance in the applicability of the method for actual watermarking, as in “proof-of-ownership watermarking”. After all, most people using someone else’s sound files in their work will transform them or add overlaid material (like a sample in a mix). So I believe that the practical scenario in terms of actual “watermarking” is precisely that of the watermarked signal being drowned into a global mix, and what to do to preserve the marking without destorting the original signal beyond measure. Obviously the case for steganography would be totally different, because then the original “carrier” information will generally be passed on unchanged. I suppose that there must be an already ample literature on that subject. I will have to look into it. Thanks for drawing my attention to that topic !
I do think this is really cool, but humans hear between 2k and 5k really well, so carving out space there should be very noticeable, actually. Wouldn't you want to do it in a way that you can't hear so much? Also I personally wouldn't want to ruin the integrity of the mono mix. Some people still listen to things in mono on phones and blutooth speakers, I wouldn't want people to assume it as a mixing decision of my art when it's not. Apjex twin did something similar, but he did it as an artistic choice. If higher samplerate audio was more common, i would prefer to just push the watermark above 20k.
I don't use FL Studio and didn't really follow the details of the compression plugins. But couldn't I just - generate an audio track that has my image as its spectrum - split the music into mid/side - subtract the image track from the mid - recombine the mid/side tracks?
thats another way to do it yeah i do subtract the image from the main signal mix that with the main signal with a polarity flip to get the part that was subtracted like a cookie cutter (i'll call it the cookie signal) i make the cookie signal mono and flip one channel then mix it back with the carved one then mix them
Listening on my phone, I can hear the watermark in the first clip, probably because my phone is mono.. (and I'm familiar with the sound of spectral images) But it's still really cool tech! I'm curious to see how it sounds through headphones, since i know inverting the phase of an entire channel makes it sound like it's coming from inside your head Have you tried listening to this through speakers? I wonder if some weird cancellation patterns would occur (i have no experience with stereo speakers)
with actual speakers its never gonna be perfectly aligned in phase so it you might head some semi-flanging effect instead, not too big of a thing but you'd definitely hear something
I don’t know who you are just stumbled upon that vid but i love your style i tried to insert | | | | | |_ in a track a few years ago you did it much better!
I am not a musician or sound designer, but this stuff is still lit AF. Didn't understand much of what was explained, but it's gotta be the coolest thing to watermark your music inside the actual music, and I guess without a proper way of erasing that watermark from the music. So the cool idea about that is, no matter where this music ends up, it will always carry it's watermark. Nice. Although if the music itself gets altered, so will the watermark I believe. But in that case, the music quality would already audidly be downgraded. I could only think of one thing to maybe keep audio quality while messing with the watermark - just getting rid of one channel, I guess at least. Getting rid of one channel would mean, there is no way of cancelling one out. Erase one channel then render in mono, should be fine I guess. Still, mono is a mentionable downgrade from Stereo, so still quite the drop in music quality.
hell yeah, no more mono compatibility one day im gonna make a song that is 100% side information, and nobody on mono will be able to hear it because its extremely funny
Alternatively instead of doing all the phase invert and mix workaround stuff, just use a mid-side matrix and run spectral compressor only on the mid channel.
oh my g-god.....;;; im s-so sorrgy,,,, i d-didtnn kn-know.......... i w-was being sO SLOW........... 😭😭💔💔💔 i j-just w-wanted to s-sjhare my p-passion;;; b-but i FAILED U.......... six wh-whole minutjes....;; an et-eternityy.......... i c-cant bbvelieve i m-made u s-suffer l-like this..........(˃̣̣̥ヘ˂̣̣̥) i-im g-gonna go r-reflect on my l-life choices in mmny cold darkj roomn,,,,, brb............
i dont care
o-oh… y-you don’t… c-care…? (╥﹏╥) i-it’s okay, i guess… i just… i w-wanted to share something i worked really hard on… b-but… i-it’s fine… i’ll just… sit here… in the d-dark… w-with my spectral w-watermarks… a-alone… (´・ω・｀) d-don’t worry about me… i’ll be okay… maybe… (っ- ‸ - ς)
Hi @@idiotinium
@@eron2761 hi
Omg bro doesn't care. Literally the Joker.
Bro cares so little that he took time out of his day to look away from his own pitifully unpopular channel and lack of subscribers to tell us about it.
insane 5 am pull
real
Fr
omg haiiii syn :33333
literally
@ratewcropolixhaii
"this sounds normal in stereo but weird when i collapse to mono"
me watching with one earbud at work and hearing the same thing twice:
wilddd
Nope that's not how that works 😅 you would have to have both earbuds in one ear (and perfectly aligned) to create the wave collapse.
No, it's still very different. In mono the waves cancel each other out. You can hear a little bit of weirdness in stereo, but that could be part of the song.
are you guys sure his phone isnt just using mono audio
guys the earbuds switch to mono audio when you only have one in it's a pretty standard feature at this point
Bazinga I'm gonna use this to hide launch codes in my videos
bazinga
bazinga
You should hide the entire script of your "fallout 3 is better than you remember is bad actually" video in your next video. its like a free video inside your video it's free real estate.
good luck
bazinga
Type of shit you rewatch everytime you're about to publish a snog
I love me a good snog
A snog
@@237vu Exactly
a snog
@@237vuI read it correctly the first time, it said song 😭
arg creators are gonna have a field day with this one
yuh, i spectrally carved morse code out of some of my songs but near 20k so you need the lossless files to see them teehee
@@idiotiniumI was thinking of using morse code in my first album. Never seen anyone do that before. I don't know if I'll use this method you just showed us but it looks interesting!
@@lilyluhtwizzyi think ive heard only 1 other artist use morse code in their albums in some way if i remember correctly so like if you do it there will be atleast 2 artists that have used morse code so like do it it would be cool
to be more specific i carved out the morse code from the spectrum at 20khz, so its entirely inaudible
They already do😂
Now I can invert the phase of all frequencies above 3000 or something hertz so it would have telephone quality in mono
oh shit thats a cool idea
Oh that’s easy, just put a lowpass on the Mid channel on the master, and make the Side channel dry.
@@zionjaymes4415 right I was like pass filters invert phase already so why go through the trouble lol
when collapsed to mono make it so you can hear someone talking diss your setup for being mono
@ LOL the side channel being so loud it completely masks a mono sample which is just "wtf u listening in mono what trash" 😆
When I was a kid in the 70's and 80's, "the church" was freaked out about subliminal satanic messages in rock-and-roll music. I played so much Led Zep, Queen and Beatles backward and learned how to "talk backward" and make my own. I had a lot of fun inserting backward-masked messages into "church" music. Now we are in the digital age, and your video has inspired me to take the trolling to the next level!!!!
Wonderful idea: backmask a bunch of satanic messages, but in the spectrum, have things like "eat your vegetables" and "don't do drugs"
thats true, though. they still do it today...
greatest troll of all time
learning to talk backwards is insane lmao
wow!!!!!! so heckin epic chungus reddit!!!!
@@Sammysapphira im gonna cry now
Aphex twin in 1999:
Ah yes of course the heisenberg uncertainty principle!
It's not as trippy as it sounds, 3blue1brown has a great video about it. Basically it just means that waves aren't instantaneous. In order to contain very specific frequencies, a signal needs to have a certain length. The resolution of the frequencies corresponds to the y resolution on the spectrogram, and the length of the samples corresponds to the x resolution. So the uncertainty principle says that higher y resolution comes at the cost of lower x resolution
heisenberg
@@cipherjoe9 what do we think of the sound of singularity
"I've been rappin' for about 17 years" - watch out for the dynamic range ahead.
say my name
As soon as OpenAI adds music generation to their model Sam Altman will be all over the podcast circuit crying about how music watermarks poison the well.
this IS genuinely the music equivalent of using that one program to "poison" images
Years ago ( like nearly 20 ), me and a friend of mine were looking to find a way to copyright music with a similar technique. Objectively, the imprint was inaudible, but would be very easy to see when looked at with the method you show. Ultimately we never took it anywhere. Two reasons for that. One was that Sony had a similar technology for their Play Station systems and DVD's. Second reason is that we didn't believe CD's would last much longer, and that most audio would go to digital format anyway, meaning all the metadata would be a part of it. Cool way to make visual sound though!
There's a spectral clipper in harmor itself btw, to clean up the initial image instead of loading an external plugin
ya but its not that great, using an external plugin processes the sound in the frequency domain instead of actually just changing the envelope of the partials
it is quite insane to think that artist could do this back in the early 2000's
i wonder how many hoops they had to jump through to achieve the same thing compared to today
Pretty cool, I wouldn’t advise doing this if you are going to play the music in a club setting. The systems are always mono.
thats what i dont get about clubs, why is it mono?
i'd understand if they setup the sub to be mono out of a single subwoofer, but the mid and high end being mono too doesnt really make sense to me
and the only stages that i've worked on actually had a stereo sub too honestly
@@idiotinium because if it’s in stereo and you’re standing close to one speaker and not the other(s), the sound becomes all skewed and wrong.
Sometimes they are in mono, yes, but definitely not always.
Agreed, they're definitely not always mono summed nowadays.
@@idiotinium Much easier to adjust volumes etc if the tracks are on a per speaker basis and mono instead of full on stereo or surround on a single track based on the club room.
The subway surfers at the end really helped with keeping my attention on the video!!! Such a lifesaver, wouldn't have made it to the end without it.
Jokes aside, this is really cool :O
i deadass didn't even see the subway surfers because the spectrogram was stimulating enough LMAO
its literally the only moving thing on the entire screen except for the aureliaPet gif
@@idiotinium ok but i didn't see it cause i was genuinely engaged in what you were saying. make of this what you will ¯\_(ツ)_/¯
@@PnfrlEnm same bruh
“Jokes aside”? You mean you don’t need subway surfers to focus on an extremely easy to digest YT video?
Idk why did youtube recommend it to me considering that I don't make music, but I have just watched the whole thing for some reason
I don’t know why’d anyone want to carve out harmonics from their track, but it’s cool!
Super creative bro, I never even thought of using harmor like this lol
It's genius
This is the type of shit I see at like 5am that renews my faith in music lol
i love that in order to draw the watermark you actually have to make a sound, it feels like using some sci-fi laser machine in an anime to stamp the watermark on the audio track.
my god, tyou just gave me a whole ass new format of communicating with people.
I actually love you, thank you so much
🙏🙏🙏🙏
First off, this is exceptional and highly-appreciated content on a subject that's needed now, more than ever.
Let me also say that I've been making and producing music (as a hobbyist) for over 2 decades at this point.
I started out on FL Studio, and have used many other DAWs, before finally settling on mainly using Ableton.
From a theoretical standpoint, I fully comprehend everything you've laid out in your thoughtful explanation.
However, I have to imagine there's a less complicated way to attain the results you're showing here, right?
The inputs are static: a 2-bit color watermark image of a specific height, and one audio file to watermark.
How is it that the process would require downloading and installing and learning and using so many tools?
I have so many things to focus on now, I shudder at the opportunity cost of setup and learning all of them.
Do I _really_ need Harmor, a spectral compressor, spectral clipping, a stereo shaper, phase inversion, etc.?
Ideally (especially with how many tracks I work on), I want to click like three things per project and be done.
harmor and AudioPaint are the only tools i know that can create a spectral image
spectral compressor is the way i know of that can compress spectrally like that which is free
the spectral gate from spectral clipping certainly helps threshold your image's signal (spectral clipper is free)
its baffling to me too that spectral tools are so obscure even tho theyre incredibly simple
Aphex twin would be proud
This is actually stupidly cool and useful. I imagine you could do this in a way that it also becomes a "signature sound" in tracks as well. Would've saved Queen some issues with Vanilla Ice lmao
"Your Honour, if it would please the court, may we switch the tracks to mono audio and review the spectrograph?"
So I could do this... Or make a "signature" sound with that "spectral stamp" and use it in every song. Or just use my producer tag in every song. Or a mixture of the three. Thanks for making this, I never imagined the sheer concept of watermarking could trigger some inspiration in me.
no way she did the tutorial
edit: thanks for correcting
she*
@@idiotiniumshe?
heee
Hey rejd
@@coulmusic wtf hi oomf
I barely understood what's happening on technical level but this looks like what you said. Clever way to sign/watermark your sounds and music. Inaudible to me after fully polished. Now draw her making garlic bread
i'll add that to the list of comissions, will get back to you eventually
Hey since you're doing the 3 thumbnail test thing, I clicked on the black thumbnail but it was the weird one with the face that caught my eye and made me add to watch later, my vote to the weird face one👍
lmaoo, yeah its a funny image to grab clicks, but the other two are more true to what is being shown in the video
I clicked on this video from the frame my patched app showed me - from 2:23
So here's a vote from me that no thumbnails are the best thumbnails cause I've never seen your channel before this vid
thank you for uploading this. iv been wanting to do something like this for years but have never been able to figure it out
heisenberg uncertainty principle located in the wild
it so insane that everytime im like messing around trying to learn how to do things like this or other weird experimental stuff and then someone uploads a video soon after
That's awesome, but in my opinion, every mix MUST be mono-compatible.
you can change this to cancel out side instead, so that its mono compatible and only shows up after inverting one channel and then collapsing to mono
to do this, swap mid and side before you send it into the chain, then swap back after (this should work, if not then idk lol)
@@idiotinium Interesting. I love the idea of embedding the watermark in the actual audio so that it can't be effectively removed once it is written.
Amazingly I just learned how to do the setup for spectrogram stuff in FL Studio so this video is gonna be a huge learning source for me!
the video i never knew i needed thank you! ✌️
This is a very neat idea. I’ve thought a lot about how you can watermark demos and instrumentals and stuff but this idea never crossed my mind. Sick tut 💪🏼
So is this how C418 made that Steve face in that one creepy music disc??
not exactly, that face hasn't been "cut out" of the spectrum, its been added in
is spectral compressor easier to install yet 😅
try this
nightly.link/robbert-vdh/nih-plug/workflows/build/master
I'm surprised at how much of this I actually did correctly when I did this. There's some things here and there that I could've done better but man was I close to getting it perfect
Thank you for explaining it
ah yes the 'say my name' uncertainty principal
Heisenberg uncertainty principle is such an insane pull that I may just have to subscribe
Fascinating. I love this syndisciplinary blend of art and algorithmic composition.
AU5 mentioned 🎉🎉🔥🔥🔥
the subway surfer footage really helped me understand everything, thanks
i'll save this for when it becomes funny enough to make spectral drawings
Sounds good until you realize every bigger stage plays your music in mono bc stereo really only works in small environments
Yup. And a good mix will be manufactured in such a way as to collapse to mono without serious side effects. I have essentially monaural hearing, so I make little combiners that feed both the left and right channels into each ear cup in my phones.
hmmm, isn't that the opposite? Smaller rooms are difficult to emulate stereo because you need to actually separate the L and R so much to have proper stereo? Don't bigger stages nowadays run proper stereo? Even decent clubs run their main system in stereo, sending mono signal only to their speakers outside the main floor (for example bar speakers)?
From what I've heard from different sound techs this is actually a myth. Properly set systems allow for impressive levels of stereo on big concerts and festivals
Bro I've been thinking with the idea to hide something in a song these days and boom your video just popped up! What a cool video ❤
your windows layout is so good i had to subscribe, also the tutorial helped
Useful. Literally will try this after watching it
I am watching this, knowing that I will never ever in my whole life will use this. Thanks I guess.
second video I've seen like this -- is RUclips tryna get me to start watermarking my music?
This is one of those irritating tutorials where you move far too fast and thus despite rewatching and refollowing, mine still doesn't work, and I find another channel that doesn't clear his throat every 3 seconds. Thanks!
just checked and i dont seem to be clearing my throat every 3 seconds
also im a she
This is incredible!
Was submitting to any% as well btw, so fun to see it from a fellow artist :D
I heard “the Heisenberg uncertainty principle” and the science nerd in me jumped. I need to know why
I did a double take and realised that watching this with one earbud is why both mono and stereo sounded the same. 😂
I appreciate the subway surfers once different window sizes and algorithms started being discussed 😭😭
Omg I completely missed that the first time lol... Seems like a clever way to throw shade
this is blessed
I wish RUclips would show me more videos like this.
Wouldnt you like to watch another short about some random hot lady doing celtic dancing where she only moves her feeeet
I'm working on a script and plugin that let you load an image, and impose it onto any existing audio! hopefully it'll make this process a lot easier.
hell yeah
yes please
I’m leaving this comment just to stay updated, I’ll buy it
@ Oh it'll be free.
someone should make a full Patcher preset that combines both the harmor patch and the compressors, that would be neat as hell
you can put generators into an effect patcher?
i forgot about that lol
If I drop an album im using this
mick gordon talked about this i remember
Pretty sure diffusion based super resolution makes these techniques pointless.
4:48 I hear this with my phone in stereo but when I put it in mono it's gone. This is so unbelievably cool.
That’s quite unexpected and fascinating. Great vid.
I will have to think of a way to use that base idea in a less destructive way (the issue cited by many commenters about large sound systems being in mono is definitely something to consider).
you can cut the image out of the side signal instead, such that people would have to flip one channel before collapsing to mono first
@ oh, that’s a great idea. A sort of double-hiding mechanism.
The thing I wonder however is how much that scrambling of phases blurs the spatial definition in the stereo mix. The exact opposite of my previous concern, so to speak ! 😁
Sorry, people are always complaining and wanting the opposite of what they have ! 🥲
This all raises a million of questions, like how it behaves if you apply the trick to only one instrument in the mix, or of other ways to embed hidden information in the phase. I can see this kind of processes be used in steganography, for instance (like is already the case in some image processing that remain invisible to the human eye).
Fascinating subject.
its not really scrambling any phases, its moreso just separating your signal into mid and size, and then you're just punching holes in the frequency spectrum. either making some regions entirely mono or entirely side (depending on what you're cutting out)
if you only apply it to one instrument then that instrument will be the only track that contains the watermark, everything else will just go over it, like if you painted a shape on a wall and then painted more shapes over it
@ I understand all this, I am just wondering how deep one may go before the hidden information get totally blurred. I guess this is important for instance in the applicability of the method for actual watermarking, as in “proof-of-ownership watermarking”. After all, most people using someone else’s sound files in their work will transform them or add overlaid material (like a sample in a mix). So I believe that the practical scenario in terms of actual “watermarking” is precisely that of the watermarked signal being drowned into a global mix, and what to do to preserve the marking without destorting the original signal beyond measure.
Obviously the case for steganography would be totally different, because then the original “carrier” information will generally be passed on unchanged.
I suppose that there must be an already ample literature on that subject. I will have to look into it.
Thanks for drawing my attention to that topic !
This is certainly neat, and i'm glad i know the concept, but I would have appreciated a slower step by step tutorial
Insanely good recommendations pull
I do think this is really cool, but humans hear between 2k and 5k really well, so carving out space there should be very noticeable, actually. Wouldn't you want to do it in a way that you can't hear so much?
Also I personally wouldn't want to ruin the integrity of the mono mix. Some people still listen to things in mono on phones and blutooth speakers, I wouldn't want people to assume it as a mixing decision of my art when it's not. Apjex twin did something similar, but he did it as an artistic choice. If higher samplerate audio was more common, i would prefer to just push the watermark above 20k.
Doing this in a very specific way might be able to futz machine learning enough to be happy. Nice.
THANK YOU, I'VE BEEN LOOKING FOR THIS ❤❤❤
You're the GOAT, not only responding through twitter but also making a WHOLE tutorial on it!!!
2:00 have you tried Harmor's spectral clipping? The slider on the top row of the interface.
Or maybe the blur knobs to declick each partial.
blur method is interesting, i might try that
ive done just the basic version of this, but holy hell what a neat way to hide it.
I am going to put this on everything even though I know nobody is going to steal my music.
going to use this and put epic poetry in all my songs
starting with 'peanus weanus ging gang geanus'
this is insane haha love it! could you do an example in other daws?
only other daw i know how to use is ableton so yea probably
I went to the doctor last Wednesday, had a bad case of spectral leakage
LMAO
Cool. Now I need someone that wants to steal the amalgamation of noises I call music.
ive thought about songs that destructively interfere when mono-ized but i never thought about using it for a watermark :0
damn, so much knowledge in so little time
Mick Gordon also did it in doom
I don't use FL Studio and didn't really follow the details of the compression plugins. But couldn't I just
- generate an audio track that has my image as its spectrum
- split the music into mid/side
- subtract the image track from the mid
- recombine the mid/side tracks?
thats another way to do it yeah
i do
subtract the image from the main signal
mix that with the main signal with a polarity flip to get the part that was subtracted like a cookie cutter (i'll call it the cookie signal)
i make the cookie signal mono and flip one channel
then mix it back with the carved one
then mix them
Listening on my phone, I can hear the watermark in the first clip, probably because my phone is mono.. (and I'm familiar with the sound of spectral images)
But it's still really cool tech! I'm curious to see how it sounds through headphones, since i know inverting the phase of an entire channel makes it sound like it's coming from inside your head
Have you tried listening to this through speakers? I wonder if some weird cancellation patterns would occur (i have no experience with stereo speakers)
with actual speakers its never gonna be perfectly aligned in phase so it you might head some semi-flanging effect instead, not too big of a thing but you'd definitely hear something
cool! you made it looks easy actually
Damn, you are very knowledgeable!!
this is my type of tutorial
I don’t know who you are just stumbled upon that vid but i love your style i tried to insert
| | |
| | |_
in a track a few years ago you did it much better!
mink gorbon talks about this in his doom 2016 gdc talk iirc
super neat!
What if you output to a analog effects and resample
it might mess with it
I am not a musician or sound designer, but this stuff is still lit AF. Didn't understand much of what was explained, but it's gotta be the coolest thing to watermark your music inside the actual music, and I guess without a proper way of erasing that watermark from the music. So the cool idea about that is, no matter where this music ends up, it will always carry it's watermark. Nice.
Although if the music itself gets altered, so will the watermark I believe. But in that case, the music quality would already audidly be downgraded. I could only think of one thing to maybe keep audio quality while messing with the watermark - just getting rid of one channel, I guess at least. Getting rid of one channel would mean, there is no way of cancelling one out. Erase one channel then render in mono, should be fine I guess. Still, mono is a mentionable downgrade from Stereo, so still quite the drop in music quality.
Most creative ways to destroy mono compatibility:
hell yeah, no more mono compatibility
one day im gonna make a song that is 100% side information, and nobody on mono will be able to hear it because its extremely funny
this feels like something i should make a pluging for
i watched this knowing i don't even make music
Dude this is actually fire
This is some nice tech man!
And so this is how Aphex Twin did it. Awesome!
i got scared when the subway surfer clip started sliding onto screen
Alternatively instead of doing all the phase invert and mix workaround stuff, just use a mid-side matrix and run spectral compressor only on the mid channel.
thats a little different but it would accomplish a something very similar, this is good too
Hurry up
oh my g-god.....;;; im s-so sorrgy,,,, i d-didtnn kn-know.......... i w-was being sO SLOW........... 😭😭💔💔💔 i j-just w-wanted to s-sjhare my p-passion;;; b-but i FAILED U.......... six wh-whole minutjes....;; an et-eternityy.......... i c-cant bbvelieve i m-made u s-suffer l-like this..........(˃̣̣̥ヘ˂̣̣̥) i-im g-gonna go r-reflect on my l-life choices in mmny cold darkj roomn,,,,, brb............
@@idiotinium ff d dfdsf
😭✋
LOL
💀