In simple layman terms, what is the Loudness War?

  • Metallica's Death Magnetic is known for having compromised sound quality in order to sound louder, we hear that it is thanks to the Loudness War.

    What is it? What impact does it have on music fans?

    I'm looking for a simple answer that can be easily understood by someone that is not familiar with audio dynamics and related concepts.

    ***Loudness war*** and other references if you search. I would do some research and ask a more specific question.

    @user3169: Almost all SE questions can get this reaction. One of the points of the network is to have answers on site, and not references to the web. I think the scope is good enough. Many people have heard about the loudness war, and an answer without prior knowledge should be good, especially on this site.

    @MeaningfulUsername I disagree about the scope. The question is answered by its first line plus "louder music grabs more attention, so track volumes have been increasing". Maybe another sentence about amplitude/frequency envelopes. To get decent information there needs to be more substance.

    @MatthewRead: The entire Wikipedia article would be a good answer here. That's not related to scope, and I don't see why there should be a lot of nitpicking on this particular questions, which is one of the few useful on the site. The expert answering the question would know what is pertinent to this issue, no?

    @MatthewRead: Why should a music fan know/care about "frequency envelopes"? I'm not sure what that means.

    @MatthewRead Your example "simple" answer is incorrect. It's a good example of why I thought having an answer carved for music fans would be helpful. Not all answers suit all askers (too simplified, too complicated, badly formulated, biased, etc) and every SE has countless answers that can be found somewhere else. Your reasoning targets most (if not all) the questions found here, music SE, and most other SE sites, so I'm not sure why an exception is being made for this question.

    This video actually explains it better than I ever could: https://www.youtube.com/watch?v=3Gmex_4hreQ

    I had a go at a user-level, minimal tech/no math explanation. Please drop in some comments if there are holes or tweaks needed in order to make it canonical. I tried to keep it light, but if I've gone under- or over-tech at any point, I can tweak.

    This whole time I thought someone was having a squabble with the 80's Metal band Loudness).

  • Tetsujin

    Tetsujin Correct answer

    7 years ago

    To explain the 'Loudness War' you need to first understand one simple fact…

    Digital audio has an absolute 'loudest noise' - it is not like old analog, where you can always 'push the fader' a bit to get some more volume.

    However you measure it, digital audio has a number of bits it can use to hold any given sample of sound. Being digital it can go as far as "all ones", it doesn't matter if it's 16-bit, 24 or 32 - the loudest any individual sample can go is "all ones" … 1111111111111111...
    To a human, it's just some numbers.

    Digital audio works down to sample level, each tiny part of the audio picture is built from an absolute number, representing a volume, or how much displacement your speaker should undergo at that precise millisecond.
    The 'sound' itself comes from each of those 'speaker positions' one after the other, so fast that we can hear the overall waveforms the speakers make in the air as the cones move in & out.

    So… if you can only make everything with an absolute maximum volume… how do you make it sound louder?

    You use compression
    Compression makes the quieter bits sounds louder - yet it leaves the loudest sounds where they are.

    Magical, yes? …
    err… No.

    When you do this you start to remove the 'punch' of those louder sounds. You can hear more sounds at louder volumes, but you still have your absolute volume limit which cannot be exceeded.
    So what starts to happen is that the 'dynamic range' - the difference between the quietest & loudest sounds - becomes smaller. For a while this just makes everything sound more 'together' - coherence of the overall instrument mix - a technique that has been used for 50 years or more, to very good effect.

    This works up to a point - & yes, it does make it start to sound 'louder'… but…

    ...after that you just squash & squash until it's all just the same volume. You can barely discern the drums are even there, because there's no longer any 'room' left for them to punch through.
    Take it to extremes & all there is is 'frequency' with no dynamic… no more loud & soft, only loud…

    Everything on 11

    We now reach the digital age, everybody wants their song to be the most noticeable on the radio. How do we do this - well, obviously we make our song louder than the others, then everybody will notice us.

    So, we compress harder

    But… radio stations already employ thousands of pounds/dollars/shekels of very finely-tweaked compression routines for their broadcast - to give their station its signature sound & keep their audiences' attention.
    see https://www.tcelectronic.com/search/Tcelectronic?text=loudness#googtrans(en|en) for just one company's range of loudness/EQ/compression-tailoring products for brodadcast radio, to see how much time & money is invested in doing this

    Apparently, that wasn't good enough - we want ours to be louder

    But… the radio stations already have…
    …but we want ours - louder…

    What actually happens eventually is that the more you compress your original track to make it 'louder' on the radio, the more the radio stations' existing compressors 'hate you'.
    Their very expensive electronics can't cope with your insistence on being 'louder' & really start to make your records sound worse than they would if you hadn't done that to them in the first place.

    … but… but… I want to be louder

    Eventually… people are learning that their records actually sound better on the radio if they actually allow them to breathe a bit …

    Frankie's "Relax" still sounds great on the radio - they had already learned in the 80s how to 'push' radio compression to sound good.

    The Eels' "Novocaine for the Soul" in the mid 90s was the first truly modern "loud" record - but they knew what they were doing & were having a private joke.

    What followed was a 'fashion war' - it could never lead to a victor; only casualties on both sides.

    The Loudness War was one that could never be won.

    The Loudness War is fortunately now seeing an armistice in the near future.
    A new standard in how we go about making our own record sound better not louder is actually on the horizon.
    Have a look at K-System metering - a relatively recent proposal, gaining credence, to balance what sounds good rather than what is loudest.

    “...it doesn't matter if it's 16-bit, 24 or 32 - fixed **or float**...” – the last one _does_ actually matter: floating point sample format lacks this hard limit. (Well, there is _a_ limit technically speaking, but not at 0 dBFS but at an absurd +768 dB (for 32-bit), which is never remotely reached in sane FP audio files.) If floating-point was standard, with _no_ official limit or even guideline for maximum-peak (and instead a requirement for audio players to normalise – hopefully, RMS rather than peak) then the loudness war would immediately be solved.

    @leftaroundabout Interesting! What's the reason of floating point sample format having such a different hard limit? Where can we learn more about that?

    None of the consumer formats use floating point so I left it out to keep things simple. On my phone, so will tweak properly when I get back

    @JCPedroza: floating point is basically the digital equivalent of Dolby noise reduction. If you directly record a highly dynamic signal on tape, the noise is very annoying in quiet parts. What Dolby does is, it strongly compresses the signal, but also stores the information about what gain reduction was used at each time. On playback, this gain is then applied back, so the quiet parts become quiet again – _together_ with the noise. In the case of floating point, the gain can also be _positive_. Peaks above 0 dB simply receive extra quantisation noise, but they aren't clipped.

    Actually, having re-read at home - I feel that the simplest 'fix' to strictly answer the question, still avoiding the maths; plus the fact that float is not in any consumer format, I'll just remove the "or float" from the answer. I fully appreciate your point but feel it would get too complex to leave it in & try explain it without 'difficult' terminology.

    Overcompression is *not* a *direct* consequence of digital recording: you can just as easily squash and brickwall a mix to death and print it to analog tape.

License under CC-BY-SA with attribution


Content dated before 7/24/2021 11:53 AM

Tags used