Re: [linux-audio-dev] Paper on dynamic range compression

Le Wed, 18 Oct 2006 13:24:55 +0100,
John Rigg <ladev@email-addr-hidden-man.co.uk> a écrit :

> On Wed, Oct 18, 2006 at 10:45:02AM +0100, Steve Harris wrote:
> > > True, but if the audio signal contains significant HF energy near
> > > the band limit, it doesn't take a very fast gain change to push it
> > > past that. Bear in mind that the ear is _very_ sensitive to aliasing
> > > artifacts, so `significant' can be a very small amount.
> >
> > These are aliasing artifacts in the sidechain though, right? So they will
> > show up as modulations in the output, rather than directly audible
> > aliasing.
>
> I was referring to aliasing in the compressed audio. Harmonic distortion
> is introduced during the attack (and decay, but usually to a lesser extent
> because it's slower).
>
> Consider a sine wave audio input. In the attack, the gain is decreasing,
> so the rising parts of each input half-cycle are reduced in slope, and
> the falling parts are increased in slope. The audio waveform starts to take
> on a sawtooth shape if the rate of gain change is high enough.
>
I cut the compressors in 2 classes: limiters as used in broadcasting
equipments, and compressors as used with electric guitars.

For the first ones, It can be true at low frequency, but most limiters will not
react faster as 1 ms or maybe 0.5 ms. And they are very simple. The analog
models are made of some kind of active component (transistor, FET, IC) acting
as a variable resistor on the signal path. The value of this variable
resistor depend on the amplitude of the signal. The signal load a simple
RC,which in turn command the value of the variable resistor. So, they are very
simple in design and it should not be so hard to develop a corresponding
algorithm. From my experience on DSP design, you should completely separate
the part calculation of the amplification factor (or value of the variable
resistor) and the signal processing. To combine the 2, you use the sound output
as source for the calculation of the amplification factor as in a real montage.

The second ones are made of some kind of active and/or passive components
(vacuum tubes, transistors, FET, IC, diodes) acting in clipping, They can
combine a limiter in the same design. But to deal only with the compressor
part, the analog clipping is not the same as the digital clipping, and I don't
see how it can be possible to reproduce it with a software without to begin
with a signal analysis of a real montage. I think at after this analysis, it
will even be necessary to process by successive approximations and trials until
the algorithm fit with the reality well enough. I know someone that done such a
software on a DSP56k, the result was great, but it take him weeks and even
months of approximations and trials to do it.

-- 
Dominique Michel