SV: Re: SV: [sdiy] Nord modular clone??

[Jeff Farr]

Nothing to do with creamware, but check out this posting by Rob Hordijk:

"Trouble with digital is that the spectrum is flat up to the limits of
hearing. That's nice for an amp and a mixer, but for several reasons
not for a synthesis device. Now analog synths are far from flat, the
spectrum bends away slightly in the high giving the high less
presence. Mostly its in the VCA and primary reason is to fight the
analog noise that filters produce. This bending is caused by odd
harmonic distortion that boosts the middle frequencies and is inherent
in most VCA circuits, figures of 3% for the typical OTA circuits used
in the VCA are common. And little bandwidth limiting condensators on
the opamps do the rest (these are often necessary to prevent the
circuits to work as a radio receiver as well). Just enough to make the
sound warm and give it more presence in the lower and middle ranges.

The effect of reducing the presence of the very high is related to
some psychoacoustic principles that have to do how the human ear
perceives spaciousness. Up to 3kHz spaciousness is perceived by the
phase and level differences between both ears. But above 3kHz the
wavelength becomes too short, shorter than the distance between both
ears. But above 3kHz another thing happens, the ear starts to work as
a combfilter that is very sensitive to the direction from where the
sound comes from. Just play one hihat sample over and over again (or
the sound of water running from a tap), close your eyes and one ear
and turn around your head. Then you hear this subtle combfilter effect
and how it tries to give you sense of direction. The ear of a healthy
person is very sensitive to recognize subtle changes and give meaning
to this information.

The question is how flat a spectrum should be. Human hearing is
evolved over much longer time than there are electronic devices. The
human ear expects to hear natural sounds, thats what it can cope with
and what it likes. Cause when one says analog sounds good, well
acoustic instruments sound even better. Just go once to a big
symphonic orchestra, middle seat, sixth row from the front and listen
to the sound of the orchestra. One may not like the music, but the
sound is just awesome, nothing like it.

The 3kHz up area has just too much presence in digital synthesizers.
Developers seem to think that because amps must be flat, virtual
analog synth sounds should be flat up to 20kHz as well. The
exaggerated presence of the very high confuses the mind when the mind
wants to combine the spatial information below 3kHz and above 3kHz
into one singular sense of direction. This even goes for a mono sound,
as the ears and mind always try to determine the direction where the
speaker is located. But due to an unbalance between the below and
above 3kHz mechanisms it appears like the mind decides the very high
is more important and tries to focus on what is happening there
instead of melting the info from both mechanisms together.

Notice that the mind is quite powerful in what it does to sound,
background sounds can be blocked completely from the awareness when
the mind has decided there is no important messages in the background
sounds. The funny thing is that when measuring the difference between
an analog synth and a digital synth the figures can be very, very
close and still one sound great and the sounds like shit. This can
only be explained by how the mind processes the audio info.

Just using an analog lowpass filter on a digital synth doesn't do the
job, as the cutoff curve is not the curve needed, it takes away the
brightness and makes the sound dull. Instead it works best to 'tilt'
the spectrum a bit. Just 1dB to 3dB per octave works wonders. Still,
you won't find an EQ on a mixing desk that does that and 1/3 octave
equalizers in general have a bad phase characteristic for this
purpose.

To clarify a bit more, listen to white noise and real pink noise. The
white noise sounds like the firebrigade directs a jet of water towards
you, while real pink noise has 'depth' and 'warmth'. So, a filter that
converts white noise to pink noise is a good candidate to use. These
filters are complex and hard to find. I used a passive one with coils
and condensators for years and right now I have an active stereo
filter that can 'tilt' the spectrum gradually between flat and 3dB per
octave with a good phase response, (must be trimmed by hand to get
both channels equal). The effect is that it does not take away any
brightness, bright sounds keep on sounding bright, but it instantly
gains presence in the middle ranges and makes sounds warmer and
immediately shifts the focus from e.g. the hats to the solo
instruments in the middle, giving them more 'body. And makes all the
digital equipment sounds much better at a straight slope of about two
dB per octave over the whole audio range. And it 'pushes' the sound a
bit back in the distance, which gives a nicer stereo field.

It is related to the high frequency damping by the atmosphere, you
probably have noticed how a thunderbolt sounds very bright without
apparent low when it strikes just a couple of meters from you, while
when it is a few miles away you only hear very low rumblings.
Apparently the atmosphere transforms the high into low over a
distance. And it is a common trick in mixing to damp the high of an
instrument to make it appear to sit more in the back. These are things
that human hearing expects by default, as ears have heard it like this
for millions of year. Its in our genes.

Well, my point is that the difference between 'analog' and 'digital'
has to do with subtle psychoacoustic phenomena that are actually known
for several decades, but are not taught in the sort of schools where
dsp programming is learned. On analog electronics it just happened to
sound good, maybe also as most of them were designed 'by ear'. Virtual
analog is apparently designed by mathematically correct models, but
those don't necessarily have to comply with psychoacoustics. Analog
equipment was designed by the same math, but it just happened that the
circuits exhibit peculiarities that just fit better with
psychoacoustics, and that when changing that one resistor it all just
sounded better. In fact, it is my strong opinion that it is how the
mind works that makes the biggest difference. I guess the best advise
is to always trust your ears more than the specs, the math and the
models. "