Recording - December 2013 - 48

like the effect on vocals, as it can hold in place a singer whose tone changes rapidly with loudness. Multiband processing, although definitely more challenging to use,
is a powerful tool.

Digital limiting
Returning to the historical overview, it was in
the 1990s, with the introduction of look-ahead
digital limiting, that the loudness wars overheated. Digital audio workstations, which allow
third party plug-ins, were quickly becoming
commonplace and a variety of new processing
tools emerged. The look-ahead limiter, first popularized in the Waves L1 plug-in, revolutionized
dynamic limiting. Ingenious in their design, lookahead limiters delay the audio path of the
processor. The control path, however, is not
delayed, and threshold is triggered and gain
reduced before the actual transient occurs in the
audio path. Figure 1 demonstrates this.
Look-ahead processing can make limiting far
more transparent, anticipating level changes
and therefore yielding greater loudness without

Louder is not better
Perhaps you have experienced the phenomenon yourself. When comparing
two audio devices, you find no issue in choosing the one that sounds better. Later,
however, you realize that one of the devices under test was in bypass, or not
patched all. You were not comparing anything, yet you had formed a decisive
opinion. What happened?
Loudness. Humans almost always think louder sounds better, just as most of us
prefer greater sweetness in a soda pop, or faster speed on a motorcycle. It is in
our basic programming and it has been proven over and over again that audio
engineers will choose processing that does nothing but make a signal slightly louder. Combine this phenomenon with digital look-ahead limiters that grant greater
loudness with less audibility and our industry faces a real dichotomy. Whereas a
traditional bus limiter might only permit 3 dB of gain reduction without side effects,
the newer processors might offer 6 or 9 dB. Why not make it louder, if you can't
tell otherwise? Louder sounds better, right?
Of course, as the cliché phrase states, there is no such thing as a free lunch.
Increasing the amount of compression applied to mix induces real side effects. One
is a loss of moment-to-moment musical dynamics. Sure, increased loudness can
bring upon some amount of visceral intensity, but lost are the differences between
soft and loud. Musical emotion has long been communicated through dynamics.
When overcompressed, everything is loud, and thus dynamics are no longer available to be used for artistic expression.

Fig 3

Humans almost always think
louder sounds better, just as
most of us prefer sweeter soda...
side effects. Soon the look-ahead limiter took
hold of the record industry, and both mix and
mastering engineers embraced the technology
to achieve greater loudness. New plug-ins continued to be marketed as well, offering better
attack and release manipulation to make compression increasingly transparent.
The introduction of look-ahead multiband
limiters was also significant. Splitting the audio
processing into multiple frequency-divided
bands, multiband processors ensure that transients in one frequency spectrum, such as the
bass-heavy kick drum, do not trigger gain
reduction in another frequency zone. The settings for each band can then be tweaked separately and/or automatically manipulated, so
optimum clarity is achieved without artifacts
such as pumping and breathing. Figures 2
and 3 demonstrate a multiband limiter.
Look-ahead multiband processing, if
designed well and employed carefully, can
provide greater mix-bus limiting without being
obvious. As a side note, it can also be used in
non-mix bus applications. Many engineers
48

RECORDING December 2013

A second issue is listener fatigue. Yes, live concerts can be very loud, but most artists
program quiet songs and other breaks into their sets to give audiences a breather
before a hard-hitting finale. Loud albums where every song is at maximum ignore the
exciting dynamics of live performances. This might even be one reason why the fulllength record is beginning to go the way of the dinosaur. Why wouldn't you put your
music library on shuffle if every song on a twelve-song album sounds the same?
Another problem with increasing loudness is that sonic artifacts emerge through
processing. Sound waves are naturally sinusoidal. Dynamically compress their
swings in amplitude and they begin to appear more like square waves. Square
waves, as anyone who as ever experimented with a synthesizer knows, contain
harmonics. Thus harmonics are generated whenever compression is applied. The
harmonics, which we call distortion, are sometimes welcome if applied subtly and
on purpose. Unfortunately, many naive engineers increase bus compression without noticing that distortion has intensified as well. It is easy to ignore the side
effects when the more accessible "louder is better" circuit is engaged in our
brains. We ignore the distortion.
Don't compress the overcompressed
The fourth issue with hypercompression is also fidelity-related. When listened to
on full-resolution 44.1 kHz/16 bit CDs, squashed recordings may have some
degree of audible distortion present. But when ripped into an MP3, AAC, or other
"data compressed" file format, overly loud audio files often see a significant
increase in distortion. That's right, dynamically-compressed recordings make bad
sounding data-compressed files.

Recording - December 2013

Table of Contents for the Digital Edition of Recording - December 2013