I was listening to a Telarc CD and read
in the notes: The signal was not passed through any processing
device (i.e. compression, limiting or equalization) during production).
They were known for their audiophile quality.
However the dynamic range of a full symphony orchestra is simply
not comfortable in the home listening environment. I realized
that what they were doing was manually riding gain or asking
the orchestra to play louder in the soft places and softer in
the loud ones. The latter procedure could be seen as interfering
with the artistic qualities of the recording.
So technically they are correct, but it
is still a form of compression.
Once upon a time I was testing a VHS recorder
and found that the input followed the outputa dB increase
in input signal caused a dB increase in output--
a threshold of 1 volt. Above that the output stayed the same
(without distortion) no matter how much the input increased,
at least up to 5 volts or so when there was heavy distortion.
This is called limiting and at least on
sine waves it works fine. With analog processing there is burst
of distortion at the onset of a loud signal. With digital systems,
including the internet, there is a delay (up to 25 seconds!)
so there is ample time to smoothly decrease the gain just ahead
of the peaks.
I looked at the waveform envelope of the
Boston Symphony broadcast stream, which sounds quite good without
obvious compression, and found that they use peak limiting.
The waveforms are normal up to a threshold, then flat-topped.
As long as the overall system gain is not too high that works
well. In the glory days of BSO broadcasts on WGBH the engineer
would follow the score and manually ride gain to bring the dynamics
within the capabilities of FM (and the analog recorder). Too
much work for producers today.
The waveform image is shown below.
email me here