Audio
Compression
By Mike Brown
Audio compression is a subject much talked about these days, but also much misunderstood.
A quick Internet search for wisdom on the subject reveals much, but almost all of it on the subject of digital data compression. Very little, however, can be learned about the subject of dynamic range reduction - the original meaning of the term.
Let us be clear. Digital
data compression concerns techniques whereby the amount of digital information
(data) required to describe and reproduce a sound can be reduced. As the amount
of data used is decreased the accuracy of the reproduced
signal decreases leading to a sound which is less detailed and may be described
as swishy, soft or grainy according to the techniques used and the severity
of the data reduction, but there is no suggestion that these techniques reduce
the
dynamic range of the sound.
Dynamic range can be
defined as the difference between the quietest and loudest sounds a system can
produce.
In life this difference is extraordinary; we talk about being able to hear a
pin drop, but compare that with the sound of a
Concorde supersonic jet taking off which is thousands of times louder, a difference
we can measure electronically
as a ratio of around 140dB. The reduction of this dynamic range is known as
audio compression.
It's done using a compressor
(or a set of them if it's stereo or greater). The device squeezes the signal
by making changes
of output voltage, which are, above a definable threshold, less than the changes
in input voltage. The degree of difference
between output and input can be set as a ratio and time constants can also be
set. Other variables can also be set and, increasingly, the sound is filtered
into a small number of frequency bands, which can be compressed individually
before
being reconstituted. This is known as multi-band compression.
The human ear works in quite an interesting way because it does, itself, compress! Exposure to very loud sounds temporarily reduces our ability to hear quiet sounds because our threshold of hearing has been shifted.
The controlled reduction
or compression of dynamic range has always played a significant part in the
professional recording process. The dynamic range of early recording systems
was inherently restricted. All analogue systems, whether mechanical
or electro-magnetic, are restricted in their dynamic range capabilities. At
the bottom end there is a level of surface
or background noise, while at the top end there is a finite maximum level imposed
either by the maximum degree of movement
in a mechanical system or voltage in an electro-magnetic system. Attempts to
reproduce a real-life dynamic range using
electronics are thus doomed to failure.
Today, the best electronic
circuits have a dynamic range capability of little more than 100dB, while the
Compact Disc
is capable of around 90dB. Classical music generally has a dynamic range of
around 50-60dB and is often limited by
the degree of ambient or background noise at the performance venue, but a good
classical recording made in a well-designed, quiet venue or studio can exceed
this and sound very exciting when played on a good system.
But some people actually
find the reproduction of such a wide dynamic very difficult to listen to. The
domestic situation rarely lends itself to the quiet study of the quietest sounds,
while the loudest can seem very anti-social.
An orchestra playing triple-forte is all very well in a concert hall but is
not always appreciated at home by the neighbours!
Pop music poses different
problems. Early pop singles were more likely to be heard on a jukebox or on
the radio than on a
high-fidelity system. As a result, artists, producers and record companies were
more interested in their record sounding loud
than anything else. So the recording engineers placed their microphones closer
to the instruments, recorded louder on tape
and mixed for maximum impact.
Pushing the electronics
and the tape up to and beyond their usual limits inevitably reduced the dynamic
range as the valves,
the tape heads and the tape itself saturated, and it was an effect which people
actually quite liked and devices were built
which took this effect further…
As used in the recording
process compression is seen (and heard!) creatively. It can be applied to individual
instruments
or to the whole mix. It can be done manually by operating a fader (usually with
reference to a musical score)
or automatically by a compressor. Either way, the result is, or should be, very
much under the control of the supervising engineer/producer.
Increasingly, over
the past 10-15 years radio stations have come to use audio compression more
and more.
At first it was applied to the output of AM radio stations in order to combat
background noise and effectively enhance their transmitter coverage area. FM
transmissions didn't suffer from inherent background noise and were chiefly
tuned to by
hi-fi buffs with good receiver and aerial systems so audio compression on FM
was seen as both unnecessary and unwanted.
It should also be pointed out that radio programmes, in the 50s, 60s and early
70s, were controlled by dedicated broadcast engineers who would work with the
presenter in order to maintain a high standard of programme output.
Effectively it was the pirate radio stations that began to put paid to this. Their one-man operations showed that the disc jockeys could effectively run the radio stations on their own without additional technical help. Of course, the pirates were AM stations where quality was far from paramount and compression was king.
It would be 10 years
or so before these practices would become widespread but nevertheless, the die
was cast.
In the autumn of 1967 the pirates were forced off the air and the BBC opened
'Wonderful' Radio 1 on 247m.
For the first time the BBC designed and built self-op studios for the disc jockeys
but the studio output was still supervised and controlled by a studio engineer,
just as it was when Capital Radio, the UK's first independent local radio station,
opened in London in 1973. Subsequent smaller stations though, were truly one-man
operated and these relied heavily
on audio compression not only to keep the station on the air safely without
the having to worry too much about his levels. Naturally presenters soon found
that if they kept their levels high the station would sound louder.
This became important
to radio stations as more and more listeners tuned away from AM in favour of
the clarity of FM.
To make matters more acute the government of the day was about to intervene.
Until the late 70s most radio stations were available simultaneously on AM and
FM. This offered a real choice: those who wished to hear radio in the background
were generally content with AM while those who sought higher quality sound could
(with the notable exception of BBC R1)
find it on FM. But once the simulcasting ended radio stations felt the need
to make the sound as compact and easy to hear
as the AM equivalent had been, particularly for those now listening to FM in
cars where reception was fragile and background noise a very real problem.
Sadly the quest for
loudness became competitive in the extreme and radio station management types
somehow came to
think that the louder their station appeared to sound the more listeners it
would attract. Quite how they arrive at that
conclusion is, frankly, baffling, but there we are. Most right thinking people
would probably consider content to be more
important than loudness!
The excessive compression
applied to most FM transmissions meant that audiophiles no longer had a source
of hi-fi radio,
but they were considered to be a very small and, frankly, cranky minority. They
are, however a minority who generally
appreciate and understand that the indiscriminate compression which radio stations
apply to their outputs does not exactly enhance the creativity of those who
make and record the music, speech and drama they may be trying to listen to.
The chance to please
audiophiles, however, returned when digital radio was introduced in 1995. Digital
Radio (DAB)
offered the broadcasters the chance to transmit the full dynamic range along
with a DRC (dynamic range control) signal
which would allow listeners the chance to opt for compression themselves if
they wanted it. Naturally this system has never
been implemented and DAB has now been degraded significantly to the point where
it is no longer considered a hi-fi medium.
It is, in a very real sense, the 21st century equivalent of AM, and intended
mainly for reception using portable receivers
which are slowly being made available and even more slowly becoming affordable.
Mike Brown is a recording engineer and also helps to edit the excellent website: http://www.transdiffusion.org