Cranked
March 17, 2015 9:29 PM Subscribe
So Google can use its powers for Good. Keep it up, boys.
posted by Johnny Wallflower at 9:40 PM on March 17, 2015 [1 favorite]
posted by Johnny Wallflower at 9:40 PM on March 17, 2015 [1 favorite]
So Google can use its powers for Good.
I'd withhold judgement on this until someone runs an analysis on the loudness of the youtube ads that play with these songs' videos.
posted by radwolf76 at 9:45 PM on March 17, 2015 [41 favorites]
I'd withhold judgement on this until someone runs an analysis on the loudness of the youtube ads that play with these songs' videos.
posted by radwolf76 at 9:45 PM on March 17, 2015 [41 favorites]
I can attest, with a decade of FM DJ experience and an FCC General Class license, that the Optimod output was always cranked i.e., music was crushed at all times.
If there was ever a one second pause between songs ("dead air" was a big no-no), you could hear the floor come up (noise) due to all that loudness/compression.
posted by CrowGoat at 10:13 PM on March 17, 2015 [5 favorites]
If there was ever a one second pause between songs ("dead air" was a big no-no), you could hear the floor come up (noise) due to all that loudness/compression.
posted by CrowGoat at 10:13 PM on March 17, 2015 [5 favorites]
If you want to know something *really* awful, I've noticed a lot of posters are uploading ripped music to YouTube where they have changed the pitch, or slightly "speeded up" the song, in order to evade content/copyright checkers. The music sound *terrible* compared to how it was originally supposed to sound, but kids listen to it anyway (kind of like scratchy AM radio, I guess, or an overdubbed cassette tape).
posted by Nevin at 10:51 PM on March 17, 2015 [5 favorites]
posted by Nevin at 10:51 PM on March 17, 2015 [5 favorites]
I'll say it again: the loudness wars are over, Merzbow won. Making something louder than noisembryo or Venereology would involve major breakthroughs in psychoacoustics and information theory.
posted by idiopath at 11:08 PM on March 17, 2015 [9 favorites]
posted by idiopath at 11:08 PM on March 17, 2015 [9 favorites]
If you want to know something *really* awful, I've noticed a lot of posters are uploading ripped music to YouTube where they have changed the pitch, or slightly "speeded up" the song, in order to evade content/copyright checkers.
I mentioned this in another thread, but those internet connected jukeboxes at bars that let you play "any song ever!" essentially do this. How much they do it varies from bar to bar, and is likely adjustable just like the sensitivity of those generally-infuriating automatic faucets.
My favorite bar to shoot pool at has it set to HILARIOUSLY sped up. Like, all the songs must be reduced to 0.75 of the typical time. It never changes the pitch, it just plays it faster pitch corrected(like what ableton live, or a lot of DJ software can do now).
So. Incredibly. Awful.
I'd withhold judgement on this until someone runs an analysis on the loudness of the youtube ads that play with these songs' videos.
I have a ginormous stereo in my office. Like, mini grateful dead wall of sound style. Gigantic under counter fridge sized sub, stacks of giant 12 and 15in woofered cerwin vega and speakerlab speakers. I like to listen to music really loud when i'm doing monotonous tasks like running DBAN on old drives that contained sensitive data. It's loud enough to knock stuff off shelves 15 feet away.
Youtube ads are so compressed. They're like, 90s fm rap station levels of compressed. I own one of those ridiculous JVC "Kaboom" boxes with the built in compressor, and it rivals the silliness of that. They're so DR crushed that they sound half the volume slider louder sometimes. I regularly hear the woofers bottom out during a stupid youtube ad and have to jump to crank the volume knob down. So dumb.
posted by emptythought at 11:20 PM on March 17, 2015 [3 favorites]
I mentioned this in another thread, but those internet connected jukeboxes at bars that let you play "any song ever!" essentially do this. How much they do it varies from bar to bar, and is likely adjustable just like the sensitivity of those generally-infuriating automatic faucets.
My favorite bar to shoot pool at has it set to HILARIOUSLY sped up. Like, all the songs must be reduced to 0.75 of the typical time. It never changes the pitch, it just plays it faster pitch corrected(like what ableton live, or a lot of DJ software can do now).
So. Incredibly. Awful.
I'd withhold judgement on this until someone runs an analysis on the loudness of the youtube ads that play with these songs' videos.
I have a ginormous stereo in my office. Like, mini grateful dead wall of sound style. Gigantic under counter fridge sized sub, stacks of giant 12 and 15in woofered cerwin vega and speakerlab speakers. I like to listen to music really loud when i'm doing monotonous tasks like running DBAN on old drives that contained sensitive data. It's loud enough to knock stuff off shelves 15 feet away.
Youtube ads are so compressed. They're like, 90s fm rap station levels of compressed. I own one of those ridiculous JVC "Kaboom" boxes with the built in compressor, and it rivals the silliness of that. They're so DR crushed that they sound half the volume slider louder sometimes. I regularly hear the woofers bottom out during a stupid youtube ad and have to jump to crank the volume knob down. So dumb.
posted by emptythought at 11:20 PM on March 17, 2015 [3 favorites]
they have changed the pitch, or slightly "speeded up" the song, in order to evade content/copyright checkers.
Just tempo or pitch shifting a song a bit shouldn't be all that audible. Something else is going on if it sounds like that.
posted by flaterik at 11:21 PM on March 17, 2015
Just tempo or pitch shifting a song a bit shouldn't be all that audible. Something else is going on if it sounds like that.
posted by flaterik at 11:21 PM on March 17, 2015
I LOVE THE TAG ON THIS POST, BTW.
posted by raihan_ at 11:37 PM on March 17, 2015 [5 favorites]
posted by raihan_ at 11:37 PM on March 17, 2015 [5 favorites]
I'd withhold judgement on this until someone runs an analysis on the loudness of the youtube ads that play with these songs' videos.
pretty much exactly. youtube lead the way on the advertising loudness war that has taken complete control of tv and web video. The reason the music videos and other videos are quiet and normalized is so that every single video on their site is quieter than their ads.
Remember: slogans notwithstanding, google is evil.
posted by shmegegge at 11:59 PM on March 17, 2015 [6 favorites]
pretty much exactly. youtube lead the way on the advertising loudness war that has taken complete control of tv and web video. The reason the music videos and other videos are quiet and normalized is so that every single video on their site is quieter than their ads.
Remember: slogans notwithstanding, google is evil.
posted by shmegegge at 11:59 PM on March 17, 2015 [6 favorites]
The author of the linked article ends by raising a few questions that he promises to answer in his next article. I wonder if there are people here who could give them a go:
The main point – YouTube is using loudness normalisation – still stands. But if you’ve been thinking ”Why are some of songs in that graph quieter or louder that -13 LUFS ?” – you’ve asked a good question.
If you’re wondering “How have YouTube implemented this ?” – you deserve a straight answer.
Actually my question is more "Why have they implemented this?"
And if you’ve spotted the big problem with the way it works, you know it needs to be discussed.
posted by rongorongo at 12:42 AM on March 18, 2015
The main point – YouTube is using loudness normalisation – still stands. But if you’ve been thinking ”Why are some of songs in that graph quieter or louder that -13 LUFS ?” – you’ve asked a good question.
If you’re wondering “How have YouTube implemented this ?” – you deserve a straight answer.
Actually my question is more "Why have they implemented this?"
And if you’ve spotted the big problem with the way it works, you know it needs to be discussed.
posted by rongorongo at 12:42 AM on March 18, 2015
one of those ridiculous JVC "Kaboom" boxes with the built in compressor
We used to use a boom box to record band practices, back in the 80's. Loud practices, with a drummer who could play at two levels: off and all the way on. It had some kind of severe limiter on the built-in mic input, and it made surprisingly clear recordings.
posted by thelonius at 12:44 AM on March 18, 2015
We used to use a boom box to record band practices, back in the 80's. Loud practices, with a drummer who could play at two levels: off and all the way on. It had some kind of severe limiter on the built-in mic input, and it made surprisingly clear recordings.
posted by thelonius at 12:44 AM on March 18, 2015
The best thing is when you are listening to an auto-compressed playlist and one of the tracks is 4'33". Not joking btw. The compressor algorithm takes the few bits of room noise or whatever down near the noise floor and pumps it up to -10db or so, such that the track becomes a long undifferentiated blast of noise.
posted by idiopath at 1:05 AM on March 18, 2015 [6 favorites]
posted by idiopath at 1:05 AM on March 18, 2015 [6 favorites]
"I LOVE THE TAG ON THIS POST, BTW."
After however many years I've been here, I've only just noticed that the mobile layout doesn't have the tags on it. I feel cheated even though I've never missed them...
posted by sodium lights the horizon at 1:22 AM on March 18, 2015 [1 favorite]
After however many years I've been here, I've only just noticed that the mobile layout doesn't have the tags on it. I feel cheated even though I've never missed them...
posted by sodium lights the horizon at 1:22 AM on March 18, 2015 [1 favorite]
Youtube ads are so compressed. They're like, 90s fm rap station levels of compressed.
I wonder, now that ads are all jumping on the banjo/accordion bandwagon, if a less compressed ad would actually be more attractive.
....and now I've given something to the ad industry. I need to go shoot myself.
posted by lumpenprole at 1:39 AM on March 18, 2015
I wonder, now that ads are all jumping on the banjo/accordion bandwagon, if a less compressed ad would actually be more attractive.
....and now I've given something to the ad industry. I need to go shoot myself.
posted by lumpenprole at 1:39 AM on March 18, 2015
I wonder if there are people here who could give them a go:
Here's my best take. I am a professional audio engineer, but I don't love loudness measurements, and am not 100% informed about loudness in general. So grain of salt, etc.
The main point – YouTube is using loudness normalisation – still stands. But if you’ve been thinking ”Why are some of songs in that graph quieter or louder that -13 LUFS ?” – you’ve asked a good question.
My guess is this - LUFS is both a point measurement and an average measurement. So if YouTube is aiming for -13 LUFS over a fairly long amount of time, that would allow for a non-zero amount of variance between different pieces of audio. It all depends on the length of the sampling window. It's also possible that some songs have enough transients left that it's not possible to boost them up to -13 LUFS without clipping (something he mentions in the article). YouTube could use a fast peak limiter to allow some more gain, but maybe they don't want to do that? I could be seriously wrong about either of those, though. I'll be interested to see the answer.
If you’re wondering “How have YouTube implemented this ?” – you deserve a straight answer.
Uuuh, what? This is just replaygain, but using LUFS as a measuring stick. They're doing a basic level analysis and adjusting output gain up or down.
Actually my question is more "Why have they implemented this?"
Unclear, but this does seem to be an industry-wide trend. This site claims that iTunes and Spotify already do this. PRSS (the Public Radio Satellite System) is rolling out LUFS in their system, in an attempt to make levels more consistent throughout programming.
And if you’ve spotted the big problem with the way it works, you know it needs to be discussed.
The coyness in this post is a little bit annoying but I'm guessing what he's talking about is what I mentioned above - it's just level riding. There's no actual change in audio content, so all of the songs are still massively compressed. They're just being turned down. The benefit of using LUFS as a measuring stick is that in the long-term, heavy-handed mastering becomes ineffective, and more dynamic songs should win out in terms of audio quality and listenability. Which will theoretically lead to a glorious golden age of lovely-sounding uncompressed music. I guess we'll see about that.
posted by god hates math at 4:08 AM on March 18, 2015 [7 favorites]
Here's my best take. I am a professional audio engineer, but I don't love loudness measurements, and am not 100% informed about loudness in general. So grain of salt, etc.
The main point – YouTube is using loudness normalisation – still stands. But if you’ve been thinking ”Why are some of songs in that graph quieter or louder that -13 LUFS ?” – you’ve asked a good question.
My guess is this - LUFS is both a point measurement and an average measurement. So if YouTube is aiming for -13 LUFS over a fairly long amount of time, that would allow for a non-zero amount of variance between different pieces of audio. It all depends on the length of the sampling window. It's also possible that some songs have enough transients left that it's not possible to boost them up to -13 LUFS without clipping (something he mentions in the article). YouTube could use a fast peak limiter to allow some more gain, but maybe they don't want to do that? I could be seriously wrong about either of those, though. I'll be interested to see the answer.
If you’re wondering “How have YouTube implemented this ?” – you deserve a straight answer.
Uuuh, what? This is just replaygain, but using LUFS as a measuring stick. They're doing a basic level analysis and adjusting output gain up or down.
Actually my question is more "Why have they implemented this?"
Unclear, but this does seem to be an industry-wide trend. This site claims that iTunes and Spotify already do this. PRSS (the Public Radio Satellite System) is rolling out LUFS in their system, in an attempt to make levels more consistent throughout programming.
And if you’ve spotted the big problem with the way it works, you know it needs to be discussed.
The coyness in this post is a little bit annoying but I'm guessing what he's talking about is what I mentioned above - it's just level riding. There's no actual change in audio content, so all of the songs are still massively compressed. They're just being turned down. The benefit of using LUFS as a measuring stick is that in the long-term, heavy-handed mastering becomes ineffective, and more dynamic songs should win out in terms of audio quality and listenability. Which will theoretically lead to a glorious golden age of lovely-sounding uncompressed music. I guess we'll see about that.
posted by god hates math at 4:08 AM on March 18, 2015 [7 favorites]
If you want to know something *really* awful, I've noticed a lot of posters are uploading ripped music to YouTube where they have changed the pitch, or slightly "speeded up" the song, in order to evade content/copyright checkers.
I think you're talking about nightcore. It does help with the automatic copyright evasion, but that was not the main motivation in its inception. It's a stylistic choice, too, a kind of style of remixing which started out as a subgenre of trance. It's totally legit. And yes, awful.
posted by procrastinator at 4:13 AM on March 18, 2015 [1 favorite]
I think you're talking about nightcore. It does help with the automatic copyright evasion, but that was not the main motivation in its inception. It's a stylistic choice, too, a kind of style of remixing which started out as a subgenre of trance. It's totally legit. And yes, awful.
posted by procrastinator at 4:13 AM on March 18, 2015 [1 favorite]
It's loud enough to knock stuff off shelves 15 feet away.
Is that not turning you deaf?
posted by bonaldi at 4:48 AM on March 18, 2015
Is that not turning you deaf?
posted by bonaldi at 4:48 AM on March 18, 2015
After however many years I've been here, I've only just noticed that the mobile layout doesn't have the tags on it. I feel cheated even though I've never missed them...
If you're in the modern theme they're down below the comments.
posted by ocherdraco at 4:50 AM on March 18, 2015
If you're in the modern theme they're down below the comments.
posted by ocherdraco at 4:50 AM on March 18, 2015
>> It's loud enough to knock stuff off shelves 15 feet away.
> Is that not turning you deaf?
What? Eh? Speak up, can't you?
posted by RedOrGreen at 7:43 AM on March 18, 2015
> Is that not turning you deaf?
What? Eh? Speak up, can't you?
posted by RedOrGreen at 7:43 AM on March 18, 2015
It had some kind of severe limiter on the built-in mic input, and it made surprisingly clear recordings.
I used to have a cheap 80's Sanyo slim box to record jams. Sounded great. I loved that thing. Best sleeper audio gear evah.
posted by ovvl at 8:46 AM on March 18, 2015
I used to have a cheap 80's Sanyo slim box to record jams. Sounded great. I loved that thing. Best sleeper audio gear evah.
posted by ovvl at 8:46 AM on March 18, 2015
I don't see how this is any sort of victory in the loudness war. Quite the opposite. It looks like songs that are already maxed out in loudness, clipped, and mutilated, are now being reprocessed to make them "quieter" which you could really only do by reducing the dynamic range even more. So if the loudness war is over, it's not because of YouTube postprocessing, it's because everyone went deaf.
posted by charlie don't surf at 9:01 AM on March 18, 2015 [3 favorites]
posted by charlie don't surf at 9:01 AM on March 18, 2015 [3 favorites]
charlie don't surf: I don't see how this is any sort of victory in the loudness war.
god hates math: The benefit of using LUFS as a measuring stick is that in the long-term, heavy-handed mastering becomes ineffective, and more dynamic songs should win out in terms of audio quality and listenability. Which will theoretically lead to a glorious golden age of lovely-sounding uncompressed music. I guess we'll see about that.
It's about what could happen, if digital audio mastering is done with YouTube as the destination, not shitty car radios and walkmen/portable CD players/cell phones amped up to 11, which is the justification for maximizing loudness in the first place (supposedly).
For a visual take on what has happened, see this snapshot of adjustments over time (Michael Jackson's "Black or White" as mastered in 1991, then remastered in 1995 and 2007; source).
If you're interested in how albums in your collection stack up, here's a Dynamic Range Database.
posted by filthy light thief at 9:14 AM on March 18, 2015 [1 favorite]
god hates math: The benefit of using LUFS as a measuring stick is that in the long-term, heavy-handed mastering becomes ineffective, and more dynamic songs should win out in terms of audio quality and listenability. Which will theoretically lead to a glorious golden age of lovely-sounding uncompressed music. I guess we'll see about that.
It's about what could happen, if digital audio mastering is done with YouTube as the destination, not shitty car radios and walkmen/portable CD players/cell phones amped up to 11, which is the justification for maximizing loudness in the first place (supposedly).
For a visual take on what has happened, see this snapshot of adjustments over time (Michael Jackson's "Black or White" as mastered in 1991, then remastered in 1995 and 2007; source).
If you're interested in how albums in your collection stack up, here's a Dynamic Range Database.
posted by filthy light thief at 9:14 AM on March 18, 2015 [1 favorite]
This is really interesting, as I have been researching and trying to find a way to help my organization (Public Access) teach audio engineering for video (broadcast and internet streaming).
The fact that there is a ton of post-processing being done to the audio levels makes getting the audio right to begin with extremely important, but also makes it so much more complicated.
The big thing I am fighting with is the very narrow dynamic range that most video broadcast specs delineate. For the most part, what I have found is that for most video production, the dynamic range is only about 20 dB, with a floor of -30dB and a peak of -10dB. The reasons for this seem to have no concrete answer, but my speculation is that by having a limited dynamic range, it forces (or allows) the content creator (or more specifically, the content distributor) to have the audio stay within a narrow range so that there are no massive jumps in the audio levels. Of course, this also means that you have to have several different versions of your video, depending on what medium you intend it to be viewed on.
One recent example I came across:
"The Devil Wears Prada" on HBA Go. My partner and I were watching this the other night, and we had to turn the volume WAY up in order to understand any of the dialogue. But then any time there was music (and there is a lot of music in that movie), it would be SO FUCKING LOUD that it scared our cats. The dynamic range that the movie was apparently mixed to would probably be fine in a theater, but on a television, hooked up to a home theater sound system, it was almost unbearable. I try to use this as an example anytime someone asks me about why we have much stricter dynamic range limits than say, a movie meant for theaters, or an online video (which tends to have a much higher peak level than broadcast video).
I am curious as to what process they are using to do this normalization, whether they are using just plain old Peak or RMS normalization, or if they are using a dynamic compressor, or even a multipressor and limiter when they are running the transcoding. That would be the technically interesting part to me.
posted by daq at 10:44 AM on March 18, 2015
The fact that there is a ton of post-processing being done to the audio levels makes getting the audio right to begin with extremely important, but also makes it so much more complicated.
The big thing I am fighting with is the very narrow dynamic range that most video broadcast specs delineate. For the most part, what I have found is that for most video production, the dynamic range is only about 20 dB, with a floor of -30dB and a peak of -10dB. The reasons for this seem to have no concrete answer, but my speculation is that by having a limited dynamic range, it forces (or allows) the content creator (or more specifically, the content distributor) to have the audio stay within a narrow range so that there are no massive jumps in the audio levels. Of course, this also means that you have to have several different versions of your video, depending on what medium you intend it to be viewed on.
One recent example I came across:
"The Devil Wears Prada" on HBA Go. My partner and I were watching this the other night, and we had to turn the volume WAY up in order to understand any of the dialogue. But then any time there was music (and there is a lot of music in that movie), it would be SO FUCKING LOUD that it scared our cats. The dynamic range that the movie was apparently mixed to would probably be fine in a theater, but on a television, hooked up to a home theater sound system, it was almost unbearable. I try to use this as an example anytime someone asks me about why we have much stricter dynamic range limits than say, a movie meant for theaters, or an online video (which tends to have a much higher peak level than broadcast video).
I am curious as to what process they are using to do this normalization, whether they are using just plain old Peak or RMS normalization, or if they are using a dynamic compressor, or even a multipressor and limiter when they are running the transcoding. That would be the technically interesting part to me.
posted by daq at 10:44 AM on March 18, 2015
yeah I am curious how they do it also...
posted by Annika Cicada at 11:21 AM on March 18, 2015
posted by Annika Cicada at 11:21 AM on March 18, 2015
It's not about loudness, it's about compression. It's not about actual loudness, (although that is the end result), but about apparent loudness. Obviously, the listener ultimately controls the volume. YouTube can't decompress recordings.
Part of the problem is that few people have hi-fi's nowadays, so music is being mixed to sound good on computer speakers and earbuds, just as (some) music was mixed on car speakers in the sixties. Heck, a lot of recordings are recorded and mixed on laptops.
But my take all along has been, "Who cares?" Little of the music I listen to has this problem. The kind of pop pabulum that gets this treatment doesn't deserve my attention, and metalheads, for all I know, may actually like it.
posted by sudon't at 11:56 AM on March 18, 2015
Part of the problem is that few people have hi-fi's nowadays, so music is being mixed to sound good on computer speakers and earbuds, just as (some) music was mixed on car speakers in the sixties. Heck, a lot of recordings are recorded and mixed on laptops.
But my take all along has been, "Who cares?" Little of the music I listen to has this problem. The kind of pop pabulum that gets this treatment doesn't deserve my attention, and metalheads, for all I know, may actually like it.
posted by sudon't at 11:56 AM on March 18, 2015
"It's not about actual loudness, (although that is the end result), but about apparent loudness."
Loudness is not the same as amplitude. Loudness is a perceptual measure, specific to humans. There is no such thing as "actual loudness" as separate from apparent.
The loudness war as usually talked about isn't strictly about "loudness", because loudness isn't just about peak and average levels, it's about frequency content (if you mix 5khz and 100hz at equal amplitude, the 5khz signal is louder, because of the bell curve of human pitch sensitivity). Though as I mentioned above, there are artists like Merzbow that tailor a track to be as absolutely loud as possible, by emphasizing frequencies that the ear is most sensitive to and keeping all levels as constantly high as the medium will allow, most people will find this "music" extremely unpleasant, if not painful to listen to at normal listening levels.
The loudness war is about keeping the amplitude in a popular music track at or near the peak possible level for the medium of distribution, so that the only difference between the "loud" and "quiet" parts of a track will typically be the timbre / texture and not the actual level of sound produced.
posted by idiopath at 12:11 PM on March 18, 2015
Loudness is not the same as amplitude. Loudness is a perceptual measure, specific to humans. There is no such thing as "actual loudness" as separate from apparent.
The loudness war as usually talked about isn't strictly about "loudness", because loudness isn't just about peak and average levels, it's about frequency content (if you mix 5khz and 100hz at equal amplitude, the 5khz signal is louder, because of the bell curve of human pitch sensitivity). Though as I mentioned above, there are artists like Merzbow that tailor a track to be as absolutely loud as possible, by emphasizing frequencies that the ear is most sensitive to and keeping all levels as constantly high as the medium will allow, most people will find this "music" extremely unpleasant, if not painful to listen to at normal listening levels.
The loudness war is about keeping the amplitude in a popular music track at or near the peak possible level for the medium of distribution, so that the only difference between the "loud" and "quiet" parts of a track will typically be the timbre / texture and not the actual level of sound produced.
posted by idiopath at 12:11 PM on March 18, 2015
There's no actual change in audio content, so all of the songs are still massively compressed. They're just being turned down. The benefit of using LUFS as a measuring stick is that in the long-term, heavy-handed mastering becomes ineffective, and more dynamic songs should win out in terms of audio quality and listenability. Which will theoretically lead to a glorious golden age of lovely-sounding uncompressed music.
It's not about loudness, it's about compression. It's not about actual loudness, (although that is the end result), but about apparent loudness. Obviously, the listener ultimately controls the volume. YouTube can't decompress recordings.
The loudness war is about keeping the amplitude in a popular music track at or near the peak possible level for the medium of distribution, so that the only difference between the "loud" and "quiet" parts of a track will typically be the timbre / texture and not the actual level of sound produced.
I'm a musician but not a sound engineer, but these comments reflect my understanding of the problem. Simply normalizing the gain of the different videos doesn't correct the problem, which is that modern music production is overly compressed and mastered to maximize the saturation at all frequencies, often to the point of inducing digital distortion into the mix. I'm looking at you, Kurt Ballou! Although he's been getting better lately. The Ghost record Infestissumam has a similar problem in the snare, despite being modeled after warm 70s heavy metal production.
If you look at the waveforms, these tracks are all set at one continuous level, like some sort of Extruded Music Product. THAT is the real problem, and is baked in at the mixing/mastering level.
posted by Existential Dread at 1:13 PM on March 18, 2015 [2 favorites]
It's not about loudness, it's about compression. It's not about actual loudness, (although that is the end result), but about apparent loudness. Obviously, the listener ultimately controls the volume. YouTube can't decompress recordings.
The loudness war is about keeping the amplitude in a popular music track at or near the peak possible level for the medium of distribution, so that the only difference between the "loud" and "quiet" parts of a track will typically be the timbre / texture and not the actual level of sound produced.
I'm a musician but not a sound engineer, but these comments reflect my understanding of the problem. Simply normalizing the gain of the different videos doesn't correct the problem, which is that modern music production is overly compressed and mastered to maximize the saturation at all frequencies, often to the point of inducing digital distortion into the mix. I'm looking at you, Kurt Ballou! Although he's been getting better lately. The Ghost record Infestissumam has a similar problem in the snare, despite being modeled after warm 70s heavy metal production.
If you look at the waveforms, these tracks are all set at one continuous level, like some sort of Extruded Music Product. THAT is the real problem, and is baked in at the mixing/mastering level.
posted by Existential Dread at 1:13 PM on March 18, 2015 [2 favorites]
This simplified waveform might help explain what has changed:
posted by Lanark at 1:55 PM on March 18, 2015 [5 favorites]
quiet: | ~~ ~~ | ~ ~ ~ ~ | ~ ~~ | |_______________ Loud Before: | ~~~~~~~~~~~~~~ | | | | |_______________ Loud after: | | | ~~~~~~~~~~~~~~ | | |_______________The loud tracks still have the compressed dynamic range (and associated crappy sound), so they are still 'loud' but that no longer gives them any volume advantage on Youtube. It will be interesting to see if the DR Database starts to show better figures coming through from new releases in the next few years.
posted by Lanark at 1:55 PM on March 18, 2015 [5 favorites]
This is the video I usually point to when I want to explain the loudness war.
posted by TwoWordReview at 2:14 PM on March 18, 2015 [1 favorite]
posted by TwoWordReview at 2:14 PM on March 18, 2015 [1 favorite]
Lanark: nice examples - here's what happened to 4'33" when it showed up on my avant garde music station (with one of these normalization / replaygain schemes implemented):
4'33" Before: | | | | | ~~~~~~~~~~~~~~ |_______________ 4'33" after: | | ~~~~~~~~~~~~~~ | | | |_______________posted by idiopath at 4:07 PM on March 18, 2015 [1 favorite]
Update to the original article.
tl;dr: current -13 LUFS target volume still way too loud but better than nothing, exact algorithm used still unknown, ads apparently also processed.
posted by Bangaioh at 6:24 AM on March 19, 2015
tl;dr: current -13 LUFS target volume still way too loud but better than nothing, exact algorithm used still unknown, ads apparently also processed.
posted by Bangaioh at 6:24 AM on March 19, 2015
I would guess that -13 LUFS is targeting mobile use cases with levelling for the benefit of play listing. If it's only newer tracks, and it isn't affecting the original track (only playback) then it's pretty likely that they're adding some sort of volume levelling metadata on upload.
posted by TwoWordReview at 9:00 AM on March 19, 2015
posted by TwoWordReview at 9:00 AM on March 19, 2015
> I think you're talking about nightcore. It does help with the automatic copyright evasion,
> but that was not the main motivation in its inception.
Which was to sound like the Chipettes, wasn't it? Because I can tell you that being a big Chipmunks/Chipettes fan segues neatly into being a big nightcore fan.
posted by jfuller at 2:38 PM on March 19, 2015
> but that was not the main motivation in its inception.
Which was to sound like the Chipettes, wasn't it? Because I can tell you that being a big Chipmunks/Chipettes fan segues neatly into being a big nightcore fan.
posted by jfuller at 2:38 PM on March 19, 2015
procrastinator: "I think you're talking about nightcore. It does help with the automatic copyright evasion, but that was not the main motivation in its inception. It's a stylistic choice, too, a kind of style of remixing which started out as a subgenre of trance. It's totally legit. And yes, awful."
I know what he's talking about. It's not nightcore, it's copyright filter avoidance.
flaterik: "Just tempo or pitch shifting a song a bit shouldn't be all that audible. Something else is going on if it sounds like that."
True, but just tempo or pitch shifting a song a bit is not enough for it to make it through the filters. You need to tweak the speed by a lot. And I know this is what's happening because almost every time I stumble across one of these tracks, there is this exchange in the comments:
Commenter: "What da fuck man this sounds like shit. Whys it so slow?"
Uploader: "Sorry, I had to slow it down cuz otherwise it would get blocked becaus of copyright."
posted by Bugbread at 8:41 PM on March 19, 2015 [1 favorite]
I know what he's talking about. It's not nightcore, it's copyright filter avoidance.
flaterik: "Just tempo or pitch shifting a song a bit shouldn't be all that audible. Something else is going on if it sounds like that."
True, but just tempo or pitch shifting a song a bit is not enough for it to make it through the filters. You need to tweak the speed by a lot. And I know this is what's happening because almost every time I stumble across one of these tracks, there is this exchange in the comments:
Commenter: "What da fuck man this sounds like shit. Whys it so slow?"
Uploader: "Sorry, I had to slow it down cuz otherwise it would get blocked becaus of copyright."
posted by Bugbread at 8:41 PM on March 19, 2015 [1 favorite]
Here's an example:
Daft Punk Random Access Memories (full album) - correct speed
Daft Punk Random Access Memories (full album) - sped up.
And, in that sped up version, this comment from the uploader: "tu eres un ignorante, si no debió subirla modificada, pero por sino le quitan el audio por el CR." (very roughly translated, "No, you're the idiot, I shouldn't have uploaded it in modified form, but if I didn't they would have stripped out the audio because of CR.")
posted by Bugbread at 11:08 PM on March 19, 2015
Daft Punk Random Access Memories (full album) - correct speed
Daft Punk Random Access Memories (full album) - sped up.
And, in that sped up version, this comment from the uploader: "tu eres un ignorante, si no debió subirla modificada, pero por sino le quitan el audio por el CR." (very roughly translated, "No, you're the idiot, I shouldn't have uploaded it in modified form, but if I didn't they would have stripped out the audio because of CR.")
posted by Bugbread at 11:08 PM on March 19, 2015
Commenter: "What da fuck man this sounds like shit. Whys it so slow?"
Uploader: "Sorry, I had to slow it down cuz otherwise it would get blocked becaus of copyright."
wait what is the fucking point of uploading the music if you have to manipulate it to the point that no one would ever want to listen to it
help me understand
posted by Existential Dread at 1:01 PM on March 22, 2015
Uploader: "Sorry, I had to slow it down cuz otherwise it would get blocked becaus of copyright."
wait what is the fucking point of uploading the music if you have to manipulate it to the point that no one would ever want to listen to it
help me understand
posted by Existential Dread at 1:01 PM on March 22, 2015
« Older The Manual of Psychedelic Support | On Memory: New Writing from Japan Newer »
This thread has been archived and is closed to new comments
Keep music dynamic.
posted by BigHeartedGuy at 9:38 PM on March 17, 2015