Re: Loudness and TV-sound?

[Daniel Levitin <levitin@xxxxxxxxxxxx>]

I'll try to answer this query.

> Fox writes: "Another technique, pioneered in the 1960s by the Tamla
>Motown studios in Detroit, uses both filters and compressors to separate
>the audible spectrum into narrow bands and pack as much energy into each
>one as possible. This equalises the level of loudness for all the backing
>instruments. The voice of one singer is then mixed to peak at a carefully
>chosen frequency so it stands out."

What Fox is describing is known as a "frequency selective compressor."  As
you described, a conventional compressor limits the dynamic range of an
input source but it does this without regard to the frequency spectrum - it
simply compresses the loudest sounds so that they are nearer in volume to
the softest sounds.  The innovation of combining frequency selectivity with
compression is that it allows the audio engineer to compress particular
bands in particular ways, allowing for much more fine tuning in the
compression process.  So for example, one could
use the equalizer section of the device to home in on the bass guitar sound
and compress it maximally, then use another equalier to home in on the
vocals and compress them maximally, and so on for various frequency bands
and instruments.  The advantage is that each signal can be compressed in a
fashion that optimizes the amount of compression that is desired for that
instrument, and the way in which the different instruments fit together to
form a desirable overall acoustic package.

The sonic result of all this compression is that it tends to "hold the
instruments together."  For example, if a vocalist is using all of her
technique to sing an emotional passage - some parts loud and some parts
soft - the soft parts can easily become lost in the band.  If you raise the
overall level of her vocal so the soft parts can be heard, the loud ones
leap out unappealingly.  Compression allows the engineer to take a
performance that spans a large range of amplitudes on the part of the
singer  and fit them into a narrower dynamic band.  This has two effects,
one practical and the other aesthetic.  First, the engineer doesn't have to
be constantly moving the fader (volume potentiometer) during the mixdown to
keep the vocal at just the right level. Second, the emotional nuances of
the singer's performance are usually retained despite the compressed
dynamic range.  This is an interesting effect; as the singer's timbre
changes during the louder passages, we hear those timbre changes without
the accompanying dynamic change.  We are all so used to it now in rock
records that they sound strange without vocal compression.

Another effect of compression is that it can "smooth out" rough
performances by blurring the attacks of instruments.  A common use is thus
to compress kick drum and bass guitar parts that were played sloppily, so
that the attack is blurred and they sound more together than they truly
are.

> Does this  technique imply
>'pushing up' levels around 3-5 kHz, where our ears are the most sensitive?
>Or do they compress the sound, say, within every 3rd octave band (or less)
>regardless of the sensitivity of our ears? Might this be related to the use
>of Exciters, that - from what I understand - are often used in TV-ads to
>generate higher intensity at the frequencies where sound seems louder?

A frequency-selective compressor can be use in any frequency range.  The
use of it doesn't "imply" pushing up levels around 3-5 kHz, but it is
certainly possible to do this.  The commercial products are not set up with
3rd octave band filters; rather, they are typically equipped with a
three-band parametric equalizer section, controlling gain, focal frequency,
and "Q" (bandwidth).  It would be hideously difficult to do a full spectrum
frequency-compression in 3rd octaves, and it would require over a dozen
separate units.  I have never heard of anyone using more than six units at
once.

>(or related to the use of psychoacoustic 'black
>boxes' like the Exciter),

The Aphex  Aural Exciter isn't a mysterious black box.  It is simply a
frequency-selective dynamic expander (the opposite of a compressor),
effective at around 10 - 16 KHz.


Daniel Levitin, Ph.D.
Stanford University Department of Music and
Interval Research Corporation
Former Recording Engineer and Production Consultant
to Steely Dan, Chris Isaak, Joe Satriani, Blue Oyster Cult