| 1.Q: |
"Why
does REQ sound different from other equalizers, such as the Waves
Q10?" |
| |
You may as well ask: "why
do any equalizers sound different from each other?".
All equalizers are based on a formula (a mathematical function
or algorithm). In hardware, this formula is implemented by
components such as tubes, transistors, integrated circuits,
resistors, capacitors, and so forth. In digital equalizers,
the algorithm works as a mathematical directly on the audio
data. Every equalizer has a different sound, with different
ranges of controls and behavior (the response of the equalizer).
Additionally, REQ is a 48-bit processor,
which increases the resolution of the equalizer and definitely
affects the sound.
|
|
|
"How
can an analog equalizer be emulated in digital processing?" |
| |
Since each equalizer, whether analog
or digital, can be essentially described as a mathematical function
(and indeed, the values of analog components are directly derived
from such math), then it is certainly possible to simply do
the same math in a digital processor. More an issue are the
curious effects of analog gear, such as extended frequency response
- with some analog equalizers going all the way to 300kHz, inductor
saturation, transformer characteristics, and so on.
|
|
|
"Why
are the bell filters of REQ wide when gain is positive, and narrower
when gain is negative?" |
| |
In a way, this is the most 'natural'
behavior of analog filters. Classic Pultec equalizers have shapes
very similar to these. By design, REQ bell filters sound and
behave closely resembling analog Pultec equalizers. For a constant
value of "Q" (width), a boost or cut will look and
sound different. These types of filters are non-symmetrical,
unlike the symmetrical filters in the Q10.
|
| 4.Q: |
"Why
do the shelving filters of REQ have a different shape?" |
| |
Primarily because they have adjustable
slope. Simply change the Q to adjust the angle of the slope
going to the shelf. Part of this type of filter, called a resonant
shelf, is the characteristic "bump" in the graph.
The overshoot/undershoot on the angle of the slope is
quite important to the sound of these shelves, first described
by Michael Gerzon in a 1994 paper.
|
| 5.Q: |
"What
does 48-bit processing really mean?" |
| |
All calculations are carried out
to a 48-bit value inside REQ. This preserves greater resolution
of fine details, and definitely affects the sound of the equalizer.
By working at this level of precision, the most exacting details
are maintained throughout the processor. Only at the output
does this 48-bit data path become reduced (to 32-bit for native
use, 24-bit for DSP use). This reduction is performed by dithering
the 48-bit internal data to the desired output.
|
