Actually, I won the Loudness War. (WORLD RECORD)
ฝัง
- เผยแพร่เมื่อ 18 ต.ค. 2024
- =Important links=
Dan Worrall's "I Won The Loudness War" video (also his other videos are pretty good): • I Won The Loudness War
ITU-R BS.1770-5 standard: www.itu.int/dm...
EBU R 128 standard: tech.ebu.ch/do...
=Sauce=
upload.wikimed...
HEYYEYAAEYAAAEYAEYAA: • HEYYEYAAEYAAAEYAEYAA
Judgment Boy from Gregory Horror Show: • Gregory Horror Show Vo...
Cromartie High School: en.wikipedia.o...
The Covenant by Droptek: / droptek-the-covenant
Background music "On Little Cat Feet" from the game OneShot: store.steampow...
Background music "Maison des fantômes" by cartonpark: cartonpark.ban...
2000's Rena dancing to 2000's trance techno: • 2000's Rena dancing to...
Yume Nikki Dream Dance: • dream dance
TF2 SFM Taunt "The Catwalk": • [SFM] Taunt: The Catwalk
Azumanga Daioh: en.wikipedia.o...
Higurashi Basketball: I have no clue but hopefully it's on archive.org somewhere
GIF of Jon Poole miming to DEVO's "Girl U Want": cdn.discordapp...
The Laughing Salesman NEW: warau-new.jp
Persimmon
The only reason I'm alive right now is due to the generous donations from patrons on Patreon, which are my only source of income. Everything I create will always be free for everybody, regardless of their financial situation, even if I'm starving to death. If you're able to help out as well, I'd be enormously grateful.
/ lostrobot
(Insert pinned comment music)
All measurements were performed at a sample rate of 44100 Hz with all samples at or below 0 dBFS.
Mistakes found so far:
1. At 4:37 I claim that a hawaiian dwarf gargling three pineapples wave would work. This is usually only the case if it only contains odd harmonics, otherwise there could be some 0 Hz DC offset present due to the aliasing. Make sure your three-pineapple-gargling hawaiian dwarves only gargle at odd harmonics relative to the precise wavelength of the recording. If your pineapple triple-gargle dwarves of hawaii clan have formed a religious pact in which all of their recordings must include even harmonics, at least ask them to ensure that those even harmonics cancel each other out perfectly when aliased to 0 Hz.
2. At 3:52 I claim that the lowpass filtered audio would play back as 0.93 dB louder on TH-cam, but I was using lossless FLAC audio when performing those measurements, which means 0.93 is incorrect. The correct way to measure this would be to first convert the audio to the same lossy audio format that TH-cam would be using, rather than measuring the lossless audio. I'm not sure what tool I'd use to get TH-cam's exact lossy audio compression, but upon testing this with 192 kbps OGG instead of FLAC, the difference was around 0.3 dB rather than 0.93 dB. The point still stands, but it's good to note that when working with lossy audio those super high frequencies are usually already filtered out.
---
dragonfruit grow on cacti and i recommend growing some because they taste good and each cactus can dump out over 200 pounds of fruit per year
its cacti not cactai poser
(this is a joke, i dont mean anything bad, i just thought being mean like that would be funny here)
@@realpv_aka_vp insert minor spelling mistake gif
Would using a higher sample rate like 96 or 192kHz (and subsequently a higher tone/Nyquist frequency) result in a higher max LUFS?
@@zacharyarney1095 Fantastic question. The answer is "almost." According to the EBU R 128 standard, Integrated LUFS should only have one digit after the decimal place. If this wasn't the case, the answer would be yes, though probably by less than 0.001 LUFS. This is solely due to the fact that the high shelf filter asymptotes to its set gain and never actually reaches it. Aside from this, there would be no difference.
@@lostrobotmusic Ahh that makes sense. I guess at 44.1 we're already at statistically insignificant increases on that logarithmic curve. Out of curiosity, did you do the math to find the lowest possible frequency at which you could achieve +0.64LUFS? From the looks of the shelf in the spec, it appears you could go down to 7kHz or maybe lower and really make some ears bleed.
Just goes to show that the real loudness was the war we made along the way
LMFAO
Madlad won the loudness war in LMMS
hahaha i'm the guy you anonymized on discord there, still appreciate you teaching me about this. always good to know what the upper boundary is for uh. idk. but you gotta know. its clearly important. you just gotta know. more LUFs = more music in a given timeframe. the most music wins. after all music is what its all about.
With this trick, I will never be the quiet kid in school again
I actually told Dan Worrall I was able to get over 6 LUFS after playing around a bit. He didn’t believe me and said it doesn’t count if I don’t publish the track. I was too lazy to do it but I’m glad you did.
I didn’t go as hight with my HPF as you did. I didn’t know over 20kHz was measured. So my track was audible but ridiculously loud
I can't believe you're making fluttershy president of the united states of america with your newly obtained power and influence
I dunno, the choice seems kind of obvious, at least to me.
@@lostrobotmusic youre so real for that
Basement dwellers rejoice *tips fedora*
@@GeneralKenobi69420wOAh!!!! Congratulations Mister GeneralKenobi69420!!!!!!!! Enjoy your very well deserved REDDIT GOLD!!!!!! And a BRAND. NEW. GOLD. FOIL. COATED. FEDORA. HOLDER.!!!!!! Congratulations!!!!!!!!!!!!!
🎉🎉🎉🎉🎉🎉🎉🙏🙏🙏🙏🙏🙏🙏👏👏👏👏👏👏👏👏👏👏👏🎉👏🎉👏🎉👏🎉🎉🔥🔥🔥
Oneshot music leading into deafening silence followed by nerding out with technicality abuse? Perfect.
(Also dissociative amnesia mood)
Ummmm sorry my meter read 999999 LUFS back in 1976 , please hand over your award
it's joever
yummers
@@KirbyCurbwhy Hey, man, I couldn't help but notice that you're drinking another milkshake today. Now I love my sweets as much as the next guy, but you've been eating too much junk food and have been gaining an unhealthy amount of weight.
this is the best format for educational content
This channel name has terrible SEO. I love it. Subbed.
My previous alias was just "Douglas", so it could certainly be worse.
The clip at the end was absolutely hilarious.
this was truly a journey
i'll now go to my kitchen to add 700 clippers with x32 oversampling to my dragonfruit cacti to squeeze the most LUFS out of it!
1:14 nice tabs
so i looked it up
its a fucking e6 rickroll
I got caught and this is the worst prank ever lmaoooo
Can't wait to EQ match my song with your audio to get that last 12dB increase I couldn't achieve.
I for one support your dragonfruit endeavors
great video with no bullshit and im now subscribed cause i like youtubers that dont make me ask more questions than get answers from
you know you're doing... something when your video has to refer to the Geneva Convention
So if I increase the sample rate to 96kbps, and play a soundwave at 48k, could I get an even higher LUFS?
Almost, if it wasn't for the fact that he EBU R 128 standard says that LUFS should only have one digit past the decimal place. Even if you had a sample rate in the quintillions and the wave playing at half that, the measurement likely wouldn't increase by more than 0.001 LU. This wouldn't end up tipping it into a score higher than the +6.4 or +6.5 measurements that were shown, which is why I didn't bother mentioning it.
Cool! *We can go higher.*
I've messed around with loudness war winning over the years, didn't know there was an unofficial competition for it. The loudest normalized file I have is +4.4 integrated, which is Dan Worrall - I Won The Loudness War but with a super crazy multiband distortion/etc patch I made to crush it as much as possible (probably improvable, it's very noisy). Later I started experimenting with using float export to push that into the hundreds of LUFS, which is like definitely cheating but still cool. My current loudest file is "no I won actually v9.wav" which uses a script I wrote to take that input audio and scale each sample to max and min double values, which is technically valid but breaks almost audio software and LUFS tester I've tried in different ways, and I haven't calculated how many LUFS that would actually be. So here's my proposal for how to win the loudness war once and for all: we take your file and run it though my script. You down?
Also yo Droptek - The Covenant, that's been in my playlist since it came out
Cool stuff! But all of this is done with the assumption that all samples are between -1 and 1, otherwise you're measuring the loudness of audio that your computer isn't even capable of playing. Even if you try to play audio outside of that range, your audio drivers will either clip it down to that range or let it wrap around within that range when converting to a lower precision.
If we were to count audio that your computer isn't actually capable of playing, the challenge would be completed by filling your hard drive with 1s and setting the first bit to 0, and the winner of the challenge would be whoever has the largest hard drive.
@@lostrobotmusic So uh, you up invading some huge server halls from governments and enormous corporations for that endeavour?
@@xNapZz why would we fill all those hard drives with 1s when we could instead fill them with pictures of cute orb-shaped fruits?
@@lostrobotmusic avocado
this is peak entertainment right here
i am the one who keeps putting grapefruits in your mailbox
wait, but i'm the one who keeps putting grapefruits in your mailbox
have we just been trading grapefruits this whole time?
@@lostrobotmusic i don't want to eat this grapefruit anymore. who knows how old it is at this point
@@ironsquid9724 I'll take it
My mail is empty and i accept grapefruits.
But it'll take months before arriving here tho
@@rossmaxx moldy grapefuir
1. the official loudness war winner, wow
2. grapefruits are in fact very tasty
Who else got confused, turned up their volume, and then got scared when he said it would damage your ears?
I predicted this, and therefore didn't actually include the audio anywhere in the video since I knew it would be inaudible anyway, knew TH-cam's lossy audio compression would almost certainly remove it anyway, and didn't want to risk damage anybody's hearing unnecessarily.
Great ending haha
Fun fact, float32 wav files can technically hold content that has 10^38 times more energy than the playback specs account for,
If volume mixing is done before that data gets sent to the sound card for whatever playback program you're using you can theoretically recover the "full" range of the file if you set the volume to 10^-38
I know the loudness war isn't about the loudest file but the loudest content, but I thought it was a interesting parallel to bring up
you also won the youtube thumbnail war
This is an incredible video. I was wondering this exact question a few months ago but didn't have the technical knowledge to answer it. Thank you for this!!
I wonder what the perceptually loudest sound at 0db peak would be. Wouldn't it be possible to get closer by using equal loudness contours instead of k-weighting to measure loudness? (For example, those defined in ISO 226:2023)
Would the loudest sound still wind up just being a sine wave at whatever frequency is considered the most "sensitive"? (Somewhere between 2-5khz probably?)
You're nearly correct. It would be a square wave (or some other wave with only full-scale values), since that adds a significant amount of power to the audio without increasing its digital peak. There wouldn't be any specific correct frequency since everybody's ears are different, but it would be around 3000 Hz or so, yes.
(A sine wave and a square wave are the same thing at Nyquist, since all of the square wave's odd harmonics would alias to the exact fundamental frequency.)
I like bees. And turtles.
I’m shocked you only have 1.2k subs! Count me in!
I like your chaotic energy
you are the perfect flavor of crazy. Amazing.
Yellow is the color of bees and bees are very good
Chad LMMS user
I would have cared, sans. I would have cared.
five star hotel mentioned!!
Im confused. Since when does a high shelf filter never reach its highest gain? Maybe Im ignorant, but setting any stock digital EQ like FF, in Bertrom curve analyzer to a high shelf does not display this behavior, and furthermore it seems to undermine my experience/conception of what a shelf filter even is. This is indeed the behavior that a tilt filter, displays but not a "regular, run of the mill" high shelf filter (though I know the two are related). In the case of FF, a high shelf filter stays consistent in its gain past 30KHz, whereas a tilt filter extends infinitely upwards
It's an asymptote: en.wikipedia.org/wiki/Asymptote
As you get higher and higher, it will infinitely get closer and closer to that gain value, but never actually reach it.
As another example of an asymptote, say you have the number 1 and it halves every second. You'd get 1, 0.5, 0.25, 0.125, 0.0625, 0.03125, and so on and so on, getting closer and closer and closer to 0 for infinity, but never actually reaching 0 no matter how ridiculously small that number gets.
The reason this isn't visible in your filter curve analyzer is because the difference is so incredibly tiny that you couldn't possibly see it, but numerically it's still there.
@@lostrobotmusic thats very interesting thanks for answering. Would a tilt filter be better suited to this application (in a theoretical sense of course)
@@Rhuggins For a loudness standard that has the same goals as LUFS, the best filter would be a specialized one that has a frequency response which approximates an equal-loudness contour: en.wikipedia.org/wiki/Equal-loudness_contour
@@lostrobotmusic thanks so much, great content!
i wanna do 10 with an uptempo track
Instantly subbed
how do we know if you haven't been playing this loudest noise the entire video? my very good ears are obviously hurting
I haven't checked, but I'm pretty sure the lossy audio codec TH-cam uses wouldn't even be able to play that audio at all.
@@lostrobotmusic yeah i was able catch that since you said it's frequency is outside of lossy compression algorithms and that there was a lossless version, so my question still stands...
how do we know if you haven't been playing this ,,, "loudest" noise the entire video? ( I do not think I can trust your word of mouth even if you say it isn't! )
If this video measured +6 LUFS TH-cam would be turning it down 20dB
@@DanWorrall the man himself
This guy takes the W
Every new producer on a pirated copy of fl studio and absolute brain dead knowledge of sound mixing. They have been doing that record for years
Lowpass one is louder on yt, eh?....
would you be able to provide a download link to the lossless file that contains the noise?
I'm worried about people accidentally playing the audio. If you have Audacity, you can go to Generate->Tone, pick "Square", Frequency of 22050 (assuming your sample rate is 44100), and Amplitude of 1.
So I definitely shouldn't download the lossless file and play it for people to damage their hearing?
You can do that with any audio if you just have a beefy enough amp and speakers for the distance and transmission losses you expect during that experiment.
Meow (epic video tbh Lost Robot)
meow :3 (thanks)
4:17 but.... but I care...
I didn't want the video to be too long, haha. I'll most likely end up making a video in the future that goes into further detail about the LUFS algorithm, as well as how it can be completely broken to trick websites into playing your audio louder than it should.
@@lostrobotmusic that’s some chaotic good I love to see! No pressure, but I would be hyped for that ^^
I've got a tune at 24 Lufs, but ok. I won't release it though because it will destroy everyones speakers
I dont believe you
Make sure you clip the audio to 0 dBFS before measurement rather than after, otherwise your measurements could go beyond thousands of LUFS because you'd be measuring audio your computer isn't even capable of playing. All samples should be within the bounds of -1 to 1.
@@lostrobotmusicI tried that,Got it to 5.9 lufs. it clearly sounded audibly quieter and not as good. Not that it sounds very nice at all, but theres a bit of a tune in that madness
wait can't you just make it a teeny tiny bit louder with a higher sample rate and a higher bitrate?
Yes, but that would most likely be a difference of less than 0.001 LU, and the EBU R 128 standard says LUFS should be rounded to one digit after the decimal.
0:10 i send my regards to your headmates
I'm not diagnosed with dissociative identity disorder, dissociative amnesia is something different: en.wikipedia.org/wiki/Dissociative_amnesia
oh sorry
@@DANKKrish No problem, it's a common mixup.
@@lostrobotmusic yeah, i used to think i only have dissociative amnesia too until shit just hit the fan and i realized that there's a bunch of other people in my head.
you should edit every video
h
ummmm my lufs goes to 11 tho
Make sure you clip the audio to 0 dBFS before measurement rather than after, otherwise your measurements could go beyond thousands of LUFS because you'd be measuring audio your computer isn't even capable of playing. All samples should be within the bounds of -1 to 1.
@@lostrobotmusic Spinal Tap joke.
Dan Worrall's challenge was mainly to see how far he could go making an actual song where all sounds are casually sitting on full scale and not exceeding it. that way he could never possibly go as high.
2:43 “plug-in I programmed by myself 🤓” “I didn’t use a libary to do this for me 🤓” “I most likely know what I’m talking about *gasps for air* 🤓”
okay miss/mister "i haven't eaten a persimmon in over three weeks because i underestimate their nutritional benefits and the resulting improvement they'd provide for my day-to-day life", whatever you say
WHERES THE FUCKING SOUND?!?
yeah, i guess you are right mathematically but this is a boring method. this is a singular frequency in unhearable range. this is just abusing the LUF definition instead of the original purpose of the loudness war which is to make music loud. Your solution is remarkable and smart but neither music nor loud (to human hearing)
Agreed, but it is genuinely interesting, I enjoyed it!
@@JoshWiniberg yes, me too
I address this in the last clip of the video. The main takeaway is that if we want to be numerically measuring for competitive loudness, we shouldn't be using something as flawed as the LUFS standard, since the competition essentially becomes "how much of this max-RMS audio can be crammed into the highest frequencies", which obviously has nothing to do with actual loudness.
Of course, if we used any other standard, I'd probably know how to break that one as well, but the point still stands.
@@lostrobotmusic ah ok, I must‘ve overheard that part, kudos to you
cool info but why does this video feel like it was edited by a schizophrenic reddit user hopped up on adderall? it's like the antithesis to how dan edits any of his videos
i was the 444th like :D
POP. FILTER.
JESUS CHRIST...
You didnt win cause its not a song
this is me when I'm loud