Perception is not reality

  • In my previous article, I made the case that all recorded music has a certain level of intrinsic production value applied to it, and that value has much more of an impact on the music’s ultimate fidelity than the sample-rate or codec in which it is delivered; two facets of digital music that typically dominate a large portion of current audiophile discourse.

    Most popular music made today has a significantly lower overall production value than that made in the past. This is largely due to the overzealous use of dynamic range compression (DRC) coupled with brickwall limiting. These techniques are employed by recording engineers at the behest of producers, labels, and even the artists themselves, to make the music sound louder. This race to the bottom is aptly named the “Loudness War” and it has been going on since the dawn of recorded music.

    The idea is that if you, the listener, hear two pieces of music back-to-back, you will always gravitate toward the louder one over the quieter one. In fact, louder sounding music initially sounds better to our ears when in reality it’s actually quite the opposite. Ergo, when volume differences present, perception is not always reality. To be a truly critical listener, you have to first check yourself (and your volume control) before you wreck yourself.


    After tens of thousands of years of evolution, the human ear has developed into a truly sensitive listening device, capable of hearing six orders of magnitude in dynamic range (140dB), three orders of magnitude in frequency (20Hz – 20kHz), and twelve orders of magnitude in sound intensity.

    And ideally you were born with two of them, allowing you to enjoy the natural world in 3-D stereo which improves your ability to localise sounds substantially over good’ole mono. Put simply, your ears allow you to eavesdrop on the softest whisper in a crowded room as well as rock out to the loudest scream at a Deathspell Omega concert. However, your hearing is not perfect, and in fact a lot of information can get lost in translation between the point in which air enters your pinna (outer ear), strikes your tymphanium (eardrum/middle ear), and is eventually converted to neural signals through the cochlea (inner ear). For example, if two frequencies close to each other are played simultaneously, there is a very good chance our ears will only detect one of them. This phenomenon is called critical masking and has to do with the fact that despite having wondrous tech specs, the inner workings of our ears still have a finite bandwidth in which they operate. Given the right auditory stimuli our perception of sound can easily be fooled. Same is true for volume.

    Volume is not linear with respect to frequency. In other words, the apparent loudness of a sound depends on its frequency and intensity. This relationship between frequency, intensity, and loudness was first studied in 1933 by Harvey Fletcher and Wilden A. Munson, who published their initial findings in their now seminal paper “Loudness, its definition, measurement and calculation.”

    By 1937, the duo had developed the first equal-loudness contour known as the Fletcher-Munson curve. It was eventually refined in 1956 and became the foundation of the official ISO 226 standard, now referred to as the Equal-Loudness contour. Loudness is measured in Phons – the official unit of measurement for volume and represents the dB SPL of a sound at a frequency of 1kHz that sounds just as loud. Our perception of volume is not a straight line across the audible frequency spectrum.


    Imagine listening to Bolt Thrower at 80dB. Your head is moving back and forth as this Brit-tank steam rolls your eardrums. Life is good. Then you turn the volume knob down and play the same song at 10dB. Take a look at the frequency disparity between the two. At 10dB, anything below 500Hz is pretty much a wash, i.e. below the threshold of audibility. Note how bass (less than 250Hz) and treble (greater than 2kHz) dramatically change in general depending on the number of phons. This graph also explains why historically speaking, most engineers tend to stay around 85dB when they mix, since that’s where the Equal-Loudness contour tends to be the flattest. It’s also why at 80dB, Bolt Thrower sounds richer and fuller than it does at 10dB since the volume across the audible frequency spectrum is more consistent.

    At this point you might be asking yourself why would I take a stand against the Loudness War? Doesn’t DRC and brickwall limiting allow engineers to make the music sound louder and therefore better? Nope.

    Between manipulating the recorded volume and adjusting playback volume there are differences. When a mastering engineer artificially pushes the volume higher by applying massive amounts of DRC, he or she is changing the recorded volume by squashing the high and low ends of the frequency spectrum. Take a look at the two masters from the track “Astrid Falls” from the last Witherscape record.


    Note: the vinyl master waveform is not from a needle drop but is sourced from the actual official digital master that was eventually pressed to wax (thanks Dan!). See all those peeks and valleys on the bottom waveform (the vinyl master)? See how they have been literally chopped off in order to raise the volume of the mid-section of the top waveform (CD master)? By simply mastering this record for maximum volume on CD its transients and imaging were substantially degraded, making the music sound lifeless and dull. More simply put, hyper compression sacrifices fidelity at the expense of volume and irrevocably damages the original recording. Read my “Take the Swanö Challenge” article for a complete lowdown (even better, take the challenge yourself).

    Now that we have a better understanding on how volume can change the way we perceive music, and how artificially raising it too high during the production process can damage it, then it should come as less of a surprise that level matching volume levels is absolutely essential when engaged in critical listening, especially when A/B-ing gear.  Otherwise, comparisons don’t enjoy a level playing field.


    How often have you stumbled across a review that claims some hot new remaster of an album you are intimately familiar with is punchier and livelier than the original (coincidentally, the new packaging and price tag for that remaster is also punchier and livelier as well)? And in that review, how often does the author explicitly state that he or she level matched the original master with the new one during their listening session? Exactly. How do you know the reviewer’s conclusion wasn’t  a result of volume differences? You don’t.

    An instant change in volume between the original and the (newly crushed) remaster can make the latter appear, at least initially, richer and fuller than the original. But once level-matched, a very different and sonically sadder story is told. The Loudness War strikes again!

    You can read more of Alex’s writings over at his Metal-Fi webzine.

    Alex M-Fi

    Written by Alex M-Fi

    Alex is co-founder and Chief Editor of, a website dedicated to the head-banging audiophile. His blood type is Type O Negative. Alex derives his income from writing software.


    1. Given what can be done now with DSP you would think that it would not be beyond someone’s capabilities to develop an after the fact filter that could restore dynamic range. It wouldn’t be perfect but I’d pay a bit to be able to optionally insert something into the chain of playback software that could offer a range of DR adjustment from subtle to aggressive etc.

      • Unfortunately, you really can’t. You can mess with phase to make the waveform *appear* more dynamic than it actually is, and of course you can always use EQ to try and bring out a bit more of a squashed kickdrum or buried cymbals.

        There is no way however to fix audible clipping caused by engineers slamming the recorded volume into the 0dBFS barrier without a limiter, or to fix the tonal damage caused by dynamic crushing. Stupidly loud albums will always sound bad. You can’t just fire up some DSP magic on Californication or Death Magnetic and expect to get decent results, or people would’ve done it already.

        The only way to fix a bad master is to go back to the original mix and do it again.

    2. Level matching should also be used while also reviewing audio hardware. And one wonders why audiophile press is an absolute joke to everyone outside the bubble.

    3. “More simply put, hyper compression sacrifices fidelity at the expense of volume and irrevocably damages the original recording”

      Very pre-Stockhausen.

      “When a mastering engineer artificially pushes the volume higher”

      You could say all recording is artificial or you could say recording is a separate thing to simply playing music.

      Have you read Perfecting Sound Forever by Greg Milner?

      • Dave, how is *HYPER* compression pre-Stockhausen? The concept of 0dbFS didn’t even exist then. I think you are confusing hyper compression (DRC coupled with brickwall limiting) with just the various methods throughout recording history to normalize volume and/or add a bit of analog fuzz (think tape compression).

        I won’t get into a pedantic discussion about the semantics of “artificial” but there is a big difference between DRC and gain.

    4. Funny, I just listened to the new Panic! at the Disco album and had to literally stop after a few minutes. It was so screechy that my ears could not handle it. Now I am beginning to understand the whole DR compression thing.

      • No surprise. Typical junk mastering. “Overs” are when the engineer didn’t even bother to use a limiter and so the track just bashes into the 0dBFS barrier and clips. Amazing that this kind of absolute trash has been business as usual since 1993.

        DR6 over -7.42 dB 01 Victorious.wav DR5 -0.11 dB -7.28 dB 02 Don’t Threaten Me With A Good Time.wav DR6 over -7.84 dB 03 Hallelujah.wav DR6 over -7.69 dB 04 Emperor’s New Clothes.wav DR4 -0.02 dB -5.74 dB 05 Death Of A Bachelor.wav DR5 over -6.07 dB 06 Crazy=Genius.wav DR6 -0.03 dB -7.26 dB 07 LA Devotee.wav DR6 -0.00 dB -7.12 dB 08 Golden Days.wav DR5 -0.00 dB -7.56 dB 09 The Good, The Bad And The Dirty.wav DR5 -0.01 dB -6.79 dB 10 House Of Memories.wav DR8 -0.20 dB -10.01 dB 11 Impossible Year.wav ———————————————————————————————- Number of files: 11 Official DR value: DR6

    5. If we’re all pretty much in agreement that this practice just leads to bad sounding music, why does it continue? Who’s benefitting? Or, who thinks they’re benefitting? I’m at the point where I won’t buy an album until I can find out the DR. So I don’t see how anyone can think that this is helping sales.

    ELAC tease Roon-enabled amplifier, speakers at CES 2016

    CES 2016: One Audio’s high-end hifi-system-in-a-box