when did music get so loud?
Monday 9/21/2026, 3:42PM
NOTE: this post was (VERY) roughly based on this other post by Novalaxia which details this process in doujin music and is actually… y’know, good, whereas this is much more focused on looking at the greater trends in music production through a technical lens, but badly. please support their work! they seem very cool and awesome
okay i swear i’ll bring something new to the table here. everyone and their grandma probably knows about the loudness war by now, and despite everyone saying it’s over, no the fuck it isn’t. people are still very much putting 20 OTTs on their masters and by god it does sound cool but stop making me use ReplayGain everytime
what’s a PCM?
so as everyone already knows, the move to digital audio with the CD brought a lot of questions to the table; how do you actually store audio digitally? the CD answers with its humble PCM format - you get 44,100 samples per second of audio, and each of those samples has a 16-bit number assigned as its loudness. higher limits are possible, but the infamous red book sticks to the above.
now, the biggest number a 16-bit integer can represent is 65,535, which means that 16-bit audio can only assign 65,535 loudness points to each sample, which infers a theoretical dynamic range of 90dB, maxing out at 0db. for listening to audio, this is more than sufficient (hell, 128kbps MP3 is transparent enough for most people) but it’s pretty suboptimal for recording, which is why most musicians record at 32-bit, or in rarer cases at 64-bit. for 24-bit audio, which is usually available with high quality lossless downloads (usually at 96kHz) the theoretical dynamic range of music amounts to 144dB, which is way beyond what the human ear can register, but sure i guess!
now, imagine a mastering engineer hears all of this and goes “what?”
this is likely because THEY ARE DEAF FROM HOW LOUD THEIR MASTERS GET
maximizing loudness (and minimizing my will to live)
during the rapid development of digital audio, engineers were faced with a dillemma; how do we make things louder? radio already applies its own compression, sure, but now, unbound from the restrictions of vinyl and its convoluted mastering process (but still bound to the restrictions of digital), we can make stuff REALLY loud.
everything i’ve set up over the course of this blog post can go to hell because the next problem becomes “how the hell do i go past 0dB.” the answer is you can’t, because if you do, you get clipping. a loud, harsh, digital distortion that makes your music genuinely unlistenable (with all due respect to modern EDM though, especially hyperpop and its derivatives, clipping is used as a stylistic choice, so i can see an argument for its use there. you don’t clip a soft rock album though)
to remedy this, we use dynamic range compression; an effect used to level out the loud and soft parts of any given audio. usually, it’s to reduce the dynamic range of said audio, which as a consequences raises the percieved loudness of that audio. the more compression you apply, the more ‘crushed’ it sounds, which can sound both awesome and horrible depending on how it’s applied. there exist who knows how many types of compressors, but the three most common ones are:
- single-band compressors: pretty self explanatory, what i just described
- multi-band compressors: these apply different amounts of compression to different frequency bands depending on how you set em up, think OTT
- limiters: a specialized compressor that prevents clipping, i.e. lowering audio that goes over the volume threshold instead of under like a standard compressor
these are most commonly used on their own, or combined into a compressor soup depending on the needs of the mixing or mastering engineer and project.
begun, the loudness wars have
and so, the new status quo was set. we starting pushing mixes as hard as possible against the hard limit, even BEYOND it sometimes with techniques like soft clipping becoming commonplace, all to increase loudness. hell, we even made LUFS, an entire new fucking unit to measure percieved loudness, because that’s how insanely loud we were getting. streaming services now have a standard for normalizing the volume of every song, because every record was mastered so wildly differently that huge jumps in volume between tracks off different albums were expected.
examples of this include Rick Rubin’s entire catalog, especially Red Hot Chili Peppers’ Californication (oh come on. you’ve already heard it), his production with System of a Down and, anecdotally, Linkin Park’s Minutes to Midnight, (What’s The Story) Morning Glory? by Oasis, and perhaps most infamously, Death Magnetic by Metallica.
Holy shit. this mastering job is so bad it made me properly capitalize a sentence for once.
this album is so delightfully horrible it became a case study for all of music production to carefully prod at with a ten foot pole, hoping to never end up making such a disaster again. the snare was crushed, the cymbals were GONE, good luck telling anything apart from anything, oh and we’re not paying your hospital bills when you eventually end up getting persistent tinnitus.
the cherry on top of the cake made of poo and dirt? these songs were featured in Guitar Hero, specifically stem-seperated unmastered versions of the songs. and GET THIS! THE INSTRUMENTS HAVE DEFINITION AGAIN!! AND THE WAVEFORM ISN’T A RECTANGLE
in the grand scheme of things, chances are that your favorite record is a brickwalled mess. and to an extent that’s ok, it’s not like you made it. but it is fun to analyze this stuff in such an aggressive manner it becomes comical, because while these practices are, objectively speaking, bad, even i enjoy putting a Maximus instance (…or three) on my drum channel to make it sound more punchy, even if it’s not really good practice.
on the topic of the music that i listen to in particular, recently i’ve been very into fade, a peculiar Japanese rock band with an American vocalist. their discography is rather conservative with compression for the most part. however, their 2011 single One Reason (the opening theme for Deadman Wonderland), which is a reworked version of the song BLACK HEARTS & DOLLAR $IGN$ from their EP Kings of Dawn is much more aggressively mastered than the original. likewise, their album released a year later in 2012, TEN (alternatively 天), is very much brickwalled, alongside not even having a lossless version on most streaming services. they, however, walked this back for their 2014 EP and greatest hits album Crossroad.
this is the part where i come to a greater conclusion about this topic and the heading is too long help
technically speaking - the loudness war isn’t over, per se. people sure as hell aren’t mastering at -14 LUFSi consistently, and streaming services normalize to their own standard anyway. most people probably don’t care, and that’s ok. hell, i even enjoy some really overcompressed albums sometimes! if you enjoy it, listen to it. the analysis is for shits and giggles, after all.
just please stop converting lossy formats to other lossy formats you’re actively making the audio sound worse how is generational loss such a complicated concept