Sunday, September 20, 2009

The Loudness Wars (aka The Volume Wars)

Recently I have totally changed my listening habits, not in terms of what I listen to but rather how I listen to it.

I usually play my library through Winamp which is a great, because highly customizable if that's what you like, free media player. I had a digital sound processor installed called Audioproc which claims to and does enhance your music through a range of compression effects, which ultimately makes your music louder as well as normalizing the volume. It was my solution to that nagging problem with MP3s whereby, because individual albums and tracks are mastered by individual mixers, you're constantly reaching for the volume knob to equalize the volume over different tracks. Audioproc also gives added overall "oomph" to your music by using, what it claims are high-quality multiband audio dynamics processors that utilize the same multiband compression/expansion/limiting technology that expensive professional broadcast hardware is based upon. 

In other words the kind of compression technology that FM broadcasters use to make their crappy 56 kbps equivalent broadcasts sound much better and louder than they otherwise would. The problem was that after a while I had a feeling I was no longer enjoying my music as much. I noticed I was getting tired of listening to my music and it all sounded flat and the same; my volume problem was fixed but now there was no dynamic range any more because of the overloaded compression processors. 

I had also noticed that more recently recorded music and MP3 downloads of old tracks tend to sound much louder than my CDs and especially vinyl from the 80s and 90s, but I thought it was just because of improved technology for a while. Then, surfing the web, I happened to come across a phenomenon called The Loudness War. Basically it's literally a war between bands and their engineers and labels to make their music sound as loud as possible in order to get an edge on the radio mainly, basically by turning up the volume as much as physically possible on the mixer without it causing obvious distortion. They use a horrible processing technique called "brickwall limiting" which actually allows for hard clipping, which means that if the signal goes over its limit it is simply clipped by allowing no more variation in the character of the output signal (what you hear). What this means is that the really loud parts of the songs sound harsh and flat. Actually, distortion is even allowed to leak through although only in very small bursts and masked by other sounds so that you, hopefully, won't notice it. (Have a listen to MGMT's Electric Feel to hear this effect in a really obvious manner. MGMT do this kind of audible clipping on purpose though - it's actually part of their sound, or so they claim.) Of course making the song loud overall reduces its dynamic range a whole lot which quickly leads to listener fatigue as a constant unvarying loud noise is unnatural and tiresome to the human ear.

I was quite shocked that audio engineers would do this just so that record companies can get a theoretical edge, therefore leading to a theoretical increase in profits at this terrible expense as regards the original quality of the music. I also realized that my own filtering of my music through Audioproc was seriously compromising the quality of what I was listening to and decided to switch it off and manually run everything through something called Replay Gain which also normalizes the volume of your MP3s but without compromising the original quality of the recording in anyway. However, it's rather tedious doing so I must admit compared to simply running a compressor.

However, doing so has made me realize I've slowly been dragged into listening to second-rate music (as far as mixing quality goes) without even realizing it due to this phenomenon of the Loudness Wars and the fact that I listen to everything on my computer or MP3 player these days. I've been paying a lot closer attention the my music and suddenly I'm seeing it everywhere, even applied to bands which otherwise have impeccable musical standards. Have a listen to the CD master of Sigur Ros's Ara Batur off their latest release for example. As the track reaches its crescendo, the pumping and breathing artefacts of over-compression become clearly audible as each instrument seems to struggle to be louder than its neighbour. And as the track features lead vocals, a choir, an orchestra with crashing cymbals as well as the band you can imagine the ugly distorted cacophony this results in. Now I realize why their latest album just didn't seem as soaring and sweeping as previous releases. Doing some visual analysis of the audio signal it appears Sigur Ros's albums are being mastered louder and louder with every new release. How can music possibly soar and sweep when it's all been expanded already and then compressed back into the box of its maximally possible limits? The whole way through? All the variation and emotion is gone; the whole work sounds dull, flat, harsh and lifeless.

It was all finally too much when I bought Shoplifters Of The World Unite And Take Over by the Smiths off the album The Sound Of The Smiths from Big Pond Music today. When I played it it just sounded, well, ugly. Much too noisy and just plain wrong. Nothing like the DVD mix I saw last night which is why I went to buy it in the first place as it wasn't in my library. I ran it through replay gain and it had to be reduced by a massive 9 dB in order to reach the reference level! That's impossibly loud for a song recorded in 1988 and released on vinyl originally. So I looked on BitTorrent for a version off the original 1987 release Louder Than Bombs and downloaded that for comparison. This time it only had to be adjusted by 0.5 dB to reach the reference level. The bass in the online store version is loud and angry and in your face, whereas in the original it's much further back in the mix, well behind the vocals and guitar. All the punch of the snare drum was gone and in the chorus the snare drum was pumping heavily as the whole signal was clipped to keep it within its limits. 

Here are waveforms of each of the two tracks just below. I reckon you can guess which one is the version affected by the Loudness War and which one is the original. These record labels and their engineers are killing our music for a quick buck. This war needs to stop.

Notice how the top waveform (which is the molested one) constantly hits the "1.0" edges of the box. These edges reflect the physical possible maximum that the wave can reach. When it exceeds this limit (which it's constantly doing) the signal has to be limited and "clipped" off, resulting in distortion. Notice in the lower original, the peaks are well within the available space for the waveform. These peaks represent the snare drum, and it's clear and has all its original punch. You can hardly make out these peaks in the molested example, and when you can they all clip. Ugly stuff.

No comments: