Kurzweil K2600 - MUSICIANS GUIDE REV A PART NUMBER 910330 CHAP... Instruction and safety manual

DSP Functions
Introduction to Algorithm Programming
16-1
Chapter 16
DSP Functions
This chapter explains the DSP functions that can be inserted into the algorithms in the Program
Editor. As you conÞgure each algorithm, the DSP functions you select determine the type of
synthesis you apply to your sounds. Deciding which algorithm to use depends on what you
want to do; thereÕs no hard and fast rule. If you want to create a classic analog sound, for
example, youÕll choose one of the algorithms containing one or more blocks that can have Þlter
functions assigned to them. If you want real-time panning effects, choose an algorithm that
includes the PANNER function in the F3 block. Your best approach is to study the algorithm
charts in the MusicianÕs Reference , and choose the algorithm that includes the functions you want
to work with.
Before we get to the explanations of the DSP functions, weÕve included a brief discussion of a
few general concepts of sound synthesis. This should help you understand the workings of the
DSP functions. WeÕll refer to these concepts repeatedly as we go along.
Any single sound waveform is composed of numerous sine wave components, each at a
different frequency. These components are called partials. The lowest frequency is perceived by
the ear as the pitch of the sound, and is called the fundamental. The other components are called
harmonics. The relative amplitudes (volume) of each of the partials in a sound determine its
timbre, its most recognizable characteristic. When you think of the difference between the sound
of a piano and a saxophone, youÕre thinking about their different timbres. A dull sound has a
strong fundamental and weak harmonics, while a bright sound has strong harmonics.
Sound synthesis can be most simply described as the manipulation of either the amplitude or
phase of one or more of the partials constituting a sound. The K2600Õs various DSP functions
give you a variety of methods for manipulating those partials. WeÕve grouped our explanations
of the DSP functions according to the types of specialized manipulation they enable you to
perform on a given sound. The categories are as follows:
Introduction to Algorithm Programming
Programming the algorithms is a multi-step process. The Þrst step is selecting an algorithm.
Changing the algorithm of an existing programÕs layer is likely to alter the sound of the layer
dramatically. As a rule, then, you wonÕt want to change a layerÕs algorithm unless youÕre
building a sound from scratch. Furthermore, when you change a layerÕs algorithm, the values
for each of the DSP functions within the algorithm may be set at nonmusical values; you should
lower the K2600Õs volume slider before changing algorithms.
Deciding which algorithm to use for a new sound is primarily a process of planning a layerÕs
signal path through the sound engine. The real sound manipulation is done by the DSP
Filters Added Waveforms
Equalization (EQ) Nonlinear Functions
Pitch / Amplitude / Pan Position Waveforms with Nonlinear Inputs
Mixers MIxers with Nonlinear Inputs
Waveforms Synchronizing (Hard Sync) Functions

16-2
DSP Functions
Introduction to Algorithm Programming
functions you insert into the algorithm. The algorithm simply lays a framework that determines
how the DSP functions interact.
Once you know which algorithm youÕre going to work with, youÕll assign various DSP
functions to each of the stages of the algorithm. These stages, as you recall, are represented by
the rectangular blocks you see on the ALG page. The arrows pointing down at the blocks
represent control inputs that affect the behavior of the DSP functions. For each arrow, thereÕs a
page of parameters controlling some aspect of the DSP functionÕs behavior. Every DSP function
has at least one control input; several have two or three.
The ALG page is where you select algorithms and assign DSP functions to the algorithmÕs
various stages. To assign a DSP function, move the cursor to select the stage you want to modify,
then use any data entry method to scroll through the list of available DSP functions for that
stage. YouÕll normally hear the effect of each selection as soon as you make it. If you donÕt hear a
difference, itÕs because the functionÕs control parameters arenÕt set to signiÞcant values. Once
you adjust some of these parameters, the function will have a noticeable effect on the sound.
Keep in mind that not all DSP functions are available at every stage of every algorithm.
When you have each stage of the current algorithm set up to your liking, you can begin to
program the control inputs of each DSP function. This is done by selecting the control-input
page(s) for the currently selected DSP function, and adjusting the parameters on the page. There
are two ways to select the control-input pages: you can move the cursor to select the DSP
function you want to tweak, and press
Edit . The selected DSP functionÕs control-input page will
appear (if itÕs a multi-stage DSP function, its Þrst control-input page will appear). Or you can
use the soft buttons to select the pages. The PITCH soft button always selects the pitch
control-input page, since the Þrst stage of every algorithm is invariably the pitch control. The
F1 Ð F4 soft buttons select the control-input pages corresponding to the remaining four arrows,
which point down at the subsequent four variable control inputs.
Figure 16-1 Input Control for DSP Functions
Each control-input page contains several parameters, which affect some aspect of the behavior
of the DSP function named on the top line of the page. Most of these parameters are the common
DSP control parameters; for a review, see Common DSP Control Parameters on page 6-14.
The possibilities are truly enormous, given the number of different combinations of functions
you can assign to any particular layer (not to mention multi-layer programs, each layer of which
has its own algorithm). You can create completely new sounds just by tweaking the parameters
on the control-input page for a single DSP function. When you begin adjusting these
EditProg:ALG|||||||||||||||<>Layer:1/1||
||||||||||||||||||||||||||||||||||||||||
||||||||||||||||||||||||||||||||||||||||
errR®rrte11231111112311111123114errR®rt|
dPITCH|gkHIFREQ|STIMULATOR|||||0kAMP||gh
CVVVVVVBCVVVVVVVVVVVVVVVVVVVVVVBCVVVVVB|
||||||||||||||||||||||||||||||||||||||||
<more||F1|FRQ|F2|DRV|F3|AMP|F4|AMP|more>
F1
FREQUENCY
CONTROL
PARAMETERS
PITCH
CONTROL
PARAMETERS
F2
DRIVE
CONTROL
PARAMETERS
F3
AMP
CONTROL
PARAMETERS
F4
FINALAMP
CONTROL
PARAMETERS

DSP Functions
Introduction to Algorithm Programming
16-3
parameters, itÕs a good idea to start with all of them set to 0 (or the value that minimizes their
effects), then adjust them one by one. This will help you understand exactly what effect each
parameter has, and will give you an idea of the variety of effects each parameter can produce.
Then you can start combining the effects of multiple parameters, and quite possibly take your
sound in a whole new direction. YouÕll quickly become familiar with the control-input pages for
the DSP functions, since most of them contain the same parameters, with just a few variations.
YouÕll Þnd that they all look much alike. The top line of each page, however, will indicate which
DSP control input it represents (PITCH, or F1ÐF4).
For example, on the page below, the top line tells you that the currently selected DSP function is
the high-frequency stimulatorÑits name is abbreviated and enclosed in parentheses. You can
also see that youÕre looking at F1, which in this case controls the frequency of the high-frequency
stimulator. So the top line of these pages always shows three things:
1. The currently selected control input (PITCH or F1ÐF4);
2. The aspect of the current DSP function controlled by the input (this varies depending on
the current DSP function);
3. The currently selected DSP function (usually abbreviated, and in parentheses). Items
1 and 2 match the label of the soft buttons that select each page. The page below, for
example, is selected with the soft button labeled
F1 FRQ
EditProg:F1|FRQ(HIFRQ|STIM)<>LAYER:1/1||
Coarse: C|4|262HZ||| |Src1||:OFF||||||||||
Fine||:0ct||||||||||Depth|:0ct||||||||||
||||||||||||||||||||Src2||:OFF||||||||||
KeyTrk:0ct/key||||||DptCtl:OFF||||||||||
VelTrk:0ct||||||||||MinDpt:0ct||||||||||
Pad||| : 0dB||| | |||||| M axDpt: 0 ct|||| | |||||
<more| | F1|FRQ | F2|DRV | F3|AMP | F4|AMP | more>
Additional Parameters
In addition to the common DSP control parameters youÕll Þnd on each page, youÕll also see a
few others on various pages. TheyÕre described here, since programming them is the same
regardless of the page on which they appear. Depending on the DSP function they affect, theyÕll
have different ranges of values and different units of measurement (%, dB, etc.).
Pad
Many of the DSP functions boost the signal as it passes through. Depending on the signalÕs
input level and the amount of gain (boost) introduced by any given DSP function, its output
may clip, which will alter the sound considerably. Clipping may also occur as a result of phase
shifting, but this is not as common as clipping caused by gain. While you may Þnd clipping to
be a useful component of some sounds, youÕll want to remove it from others. The Pad
parameter, which appears on the control-input pages of many DSP functions, lets you attenuate
(reduce the amplitude of) the signal by 6, 12, or 18 dB at the input of those functions. Use the
Pad parameter to reduce or eliminate any undesired clipping caused by the currently selected
DSP function.

16-4
DSP Functions
Introduction to Algorithm Programming
KeyTrack Start (KStart)
This parameter appears on many control-input pages, and gives you added control over the
effect of key tracking. For each note you play, it multiplies the value of the KeyTrk parameter by
a number that varies with the noteÕs MIDI key number. If KeyTrk is set to 0 , this parameter will
have no effect. When KeyTrk is a nonzero value, KStart will modify the normal key tracking
curve, which is shown is the diagram below. The effect of normal key tracking reaches its
minimum at C -1, and its maximum at C 9. You can use KStart to dampen the effects of key
tracking at one end of the keyboard. If key tracking causes a sound to clip or distort toward the
high end of the keyboard, for example, you can use KStart to reduce the effect of the key
tracking at the upper end without changing its effect on the lower end. To do this you would set
a negative value for KeyTrk, and a unipolar value for KStart.
Unipolar Keystart
The range of values for KStart is C 1 to C 9 unipolar, and C -1 to C 9 bipolar. Unipolar and
bipolar values have different effects on the key tracking. The next three diagrams illustrate the
effect of three different unipolar keystart values on the key tracking curve when a positive value
is assigned for the KeyTrk parameter. At a KStart value of C 4 , for example, there is no key
tracking effect below Middle C (it multiplies the key tracking amount by a key number value
of 0 ). The key tracking value is multiplied by 0 at C 4 (normal key tracking), by 1 at C
#
4, by 2 at
D 4, and so on to a maximum of 64 at 5 1/3 octaves above the KStart value. For higher notes, the
key tracking effect would still increase on its own, but the effect of keystart would not increase
further. At a KStart value of C 3, the key tracking value would be multiplied by 0 for C 3 and all
notes below, by 1 for C
#
3, and so on. The key number value would reach its maximum of 64
before reaching C 9. When KStart is set above C 4 , its effect on key tracking will continue to
increase up to C 9, but will not reach full scale at C 9.
YouÕll use unipolar values for KStart when you want to cancel the key tracking effect on a DSP
function over a sizable portion of the keyboard, but have it increase or decrease throughout the
rest of the keyboardÕs range. Set high unipolar values for KStart when you want to remove key
tracking from the lower notes, applying it only to the higher notes. If you have set a positive
value for KeyTrk, set low unipolar values when you want to apply key tracking to the lower
notes and pin it at its maximum throughout the upper range of the keyboard. You may want to
use low values of key tracking in this case, depending on the DSP function youÕre applying.
When the value of the KeyTrk parameter is negative, remember that the key tracking is at its
minimum effect at C 9, and maximum at C -1. In this case, the key tracking effect will be reduced
for notes above the KStart setting. For notes below the keystart note, the normal key tracking
amount will apply.
Normal KeyTrk curve
(Positive KeyTrk value)
C 4C 3 C 5

DSP Functions
Introduction to Algorithm Programming
16-5
Figure 16-2 Unipolar Keystart
Bipolar Keystart
For bipolar KStart values with positive key tracking values, the effect on key tracking depends
on whether the KStart parameter is set above or below C 4 . When itÕs above, the effect on key
tracking will be minimum at C -1 , reaching its maximum effect on key tracking at the keystart
setting. The normal key tracking curve applies above the keystart setting. When KStart is set
below C 4 , the effect on key tracking is maximum at C 9, decreasing with each successive note
closer to the keystart setting, and remaining constant at the keystart setting and below. The
normal key tracking curve applies below the keystart setting.
Figure 16-3 Bipolar Keystart
Negative KeyTrk value with
KStart value at C4 Unipolar
C 4C 3 C 5
Positive KeyTrk value with
KStart value at C 3 Unipolar
C 4C 3 C 5
Positive KeyTrk value with
KStart value at C 5 Unipolar
C 4C 3 C 5
Positive KeyTrk value with
KStart value at C4 Unipolar
C 4C 3 C 5
Positive KeyTrk value with
KStart value at C 5 Bipolar
C 4C 3 C 5
Positive KeyTrk value with
KStart value at C 3 Bipolar
C 4C 3 C 5

16-6
DSP Functions
The DSP Functions
Use bipolar settings for KStart when you want to gradually increase or decrease the key tracking
effect of the currently selected DSP function across the entire keyboard range. With KStart at C 4
bipolar, playing C 4 will apply the DSP function at the level you set with the Adjust parameter,
and will increase or decrease with higher or lower notes, depending on your settings for KeyTrk.
When KeyTrk is set to a negative value, the effect on key tracking is reversed. For keystart
settings above C 4 , the effect on key tracking will be maximum at C -1, decreasing with each
note closer to the keystart setting, and remaining constant at and above the keystart setting. For
keystart settings below C 4 , the effect on key tracking will be minimum at C 9, increasing with
each note closer to the keystart setting, and remaining constant for notes at and below the
keystart setting.
KStart is available for many of the nonlinear DSP functions , like SHAPER and WRAP. If you
like the control it gives you, you can simulate its effect by using the FUNs. To simulate unipolar
keystart, assign Key Number (KeyNum) as one of the inputs to a FUN, and select one of the
diode equations for the FUNÕs Function parameter. To simulate bipolar keystart, assign Bipolar
Key Number (BKeyNum) as one of the inputs of a FUN. Then assign those FUNs to some other
control-source parameter.
The DSP Functions
Filters
Filters are widely used in synthesis to change the timbre of a sound by manipulating the
amplitude of speciÞc partials. When using Þlters, you always set a reference point (cutoff or
center frequency) that determines which partials the Þlters affect. HereÕs a quick summary of the
effects of the Þlter functions.
Lowpass Þlters cut the levels of all partials above the cutoff frequency without affecting the
partials at or below the cutoff frequency (the low frequencies pass through). Highpass Þlters do
the opposite; they cut the levels of all partials below the cutoff frequency without affecting the
partials at or above the cutoff frequency.
Notch Þlters, as the name implies, cut the levels of partials in a range between high and low
frequency. Consequently the ÒcutoffÓ frequency is referred to as the center frequency. With notch
Þlters, the levels of partials at the center frequency are cut, while the levels of partials above and
below the center frequency are unaffected. Bandpass Þlters are the opposite of notch Þlters; they
leave the levels of partials at the center frequency unchanged, and cut the levels of partials
above and below the center frequency.
One-pole Lowpass One-pole Allpass
Two-pole Lowpass Two-pole Allpass
Two-pole Lowpass, -6 dB resonance Two-pole Notch
Two-pole Lowpass, +12 dB resonance Two-pole Notch, fixed width
Four-pole Lowpass with separation Double Notch with separation
Gated Lowpass Two-pole Bandpass
One-pole Highpass Two-pole Bandpass, fixed width
Two-pole Highpass Twin Peaks Bandpass
Four-pole Highpass with separation

DSP Functions
The DSP Functions
16-7
The use of lowpass, highpass, notch, and bandpass Þlters is often referred to as subtractive
synthesis, since the timbre of a sound is changed by removing certain partials.
Allpass Þlters, instead of cutting or boosting the partials of a sound, change the phase of the
partials as their frequencies pass through the center frequency.
FilterTerminology
Rolloff Filters do not usually cut all frequencies precisely at their cutoff point. Instead, the
amplitude of the frequencies above (or below, in case of a hi pass Þlter) the cutoff
decrease by a Þxed amount per octaveÑfor example, 6 dB per octave. This curve
of lessening amplitude is called a rolloff.
Poles The number of poles in a Þlter affect how sharp the rolloff is. The more poles there
are, the sharper the rolloff, meaning that the cutoff will have a more dramatic
effect on the sound. The K2600 has one-pole, two-pole, and four-pole Þlters
available. A one-pole Þlter has a 6 dB per octave cutoff; a two-pole is 12 dB per
octave; and a four-pole is 24 dB per octave.
Resonance In a Þlter that has resonance, the frequencies near the cutoff are given an increase
or decrease in amplitude. If you decrease these frequencies, you are essentially
creating a longer rolloff. But if you increase those frequencies thereby
emphasizing them, it creates a distinctive sound that you will very likely
recognize. Resonance is also sometimes called Emphasis or Q on various
synthesizers. Resonance on the K2600 is implemented in one of two ways. On
some Þlters, the resonance is Þxed, adding or subtracting a speciÞc amount of dB
to the affected frequencies (the ones near the cutoff). On other Þlters, you can
control the amount of resonance applied. In the case of these Þlters, there will
always be a separate control page for the resonance.
Separation Four of the Þlters in the K2600 (both Four-Pole Þlters, the Double Notch, and the
Twin Peaks) are actually two Þlters combined into one DSP function. For these
Þlters, you will Þnd a control page called Separation. This allows you to shift the
cutoff frequency of the second Þlter, creating a separation in the cutoff frequencies
of the two Þlters. In the case of the Notch and Band Pass Þlters, this can be used to
create two separate notches or band passes. In the case of the four-pole Þlters, it
affects the shape of the roll off. For the four-pole Þlters, separation set to 0 creates
sharp rolloff of 24dB per octave.
How to Read the Graphs
The graphs show the rolloff curve, using several different values to show how they change the
shape of the curve. Amplitude is always on the vertical axis. Frequency is always on the
horizontal axis. You will notice on several graphs that the curve becomes more dramatic as the
cutoff frequency is set at a higher value. This is because the highest frequency the K2600 can
produce is 20Khz, so as the cutoff is set to higher values, there are fewer frequencies available
before it is past the range of the K2600.

16-8
DSP Functions
The DSP Functions
One-pole Lowpass Filter (LOPASS)
Frequencies below the cutoff frequency are unaffected by this Þlter. At the cutoff frequency, the
signal is attenuated 3 dB. ThereÕs a rolloff of 6 dB per octave above the cutoff frequencyÑthat is,
the signal is attenuated 6 dB with each octave above the cutoff. The resonanceÑthe amount of
cut or boost at the cutoff frequencyÑis Þxed at -3dB. When the cutoff frequency is well below
the lowest-frequency partials of a sound, lowering the cutoff further will not affect the timbre of
the sound, but will reduce its overall amplitude.
EditProg:F1|FRQ(LOPASS)||||<>Layer:1/1||
Coarse: C|4|262|Hz|| |Src1||:OFF||||||||||
Fine||:0ct||||||||||Depth|:0ct||||||||||
||||||||||||||||||||Src2||:OFF||||||||||
KeyTrk:0ct/key||||||DptCtl:OFF||||||||||
VelTrk:0ct||||||||||MinDpt:0ct||||||||||
Pad|||:0dB||||||||||MaxDpt:0ct||||||||||
<more||F1|FRQ|F2|||||F3|||||F4|AMP|more>
Parameter Range of Values
Coarse Adjust C 0 16 Hz to G 10 25088 Hz
Fine Adjust ± 100 cents
Key Tracking ± 250 cents per key
Velocity Tracking ± 10800 cents
Pad 0, 6, 12, 18 dB
Source 1 Control Source list
Source 1 Depth ± 10800 cents
Source 2 Control Source list
Source 2 Depth Control Control Source list
Minimum Depth, Source 2 ± 10800 cents
Maximum Depth, Source 2 ± 10800 cents
-60
-50
-40
-30
-20
-10
0
10 100 1000 10000 100000
C5 C10
Frequency in Hertz
Amplitude in dB
Cutoff Frequency
from C5 to C10

DSP Functions
The DSP Functions
16-9
The Coarse Adjust parameter sets the cutoff frequency in terms of a key name. The remaining
parameters (except Pad) alter the cutoff frequency in increments of cents. YouÕll notice that
positive values for key tracking have an interesting effect on the function of lowpass Þlters;
positive key tracking values raise the cutoff frequency for high notes and lower it for low notes.
More speciÞcally, a value of 100 cents per key on this page, when Þltering a constant waveform
like a sawtooth, would result in waveforms of exactly the same shape for all pitches of the
waveform. The cutoff frequency moves in sync with the frequencies of the waveformÕs partials
as different pitches are generated. Negative key tracking values will steepen the rolloff of
lowpass Þlters above the cutoff. The Pad parameter, as always, attenuates the signal at the input
to the function. These parameters affect all the lowpass Þlters similarly.
Two-pole Lowpass Filter (2POLE LOWPASS)
The two-pole lowpass Þlter has a rolloff of 12 dB per octave above the cutoff frequency. This is a
two-stage function, so it has two control-input pages. The Þrst affects the cutoff frequency, and
has the same parameters as the one-pole lowpass. The second control-input page (F2 RES)
affects the resonance of the Þlter. Resonance is a cut or boost in amplitude of the partials in the
vicinity of the cutoff frequency.
Set the resonance with the Adjust parameter on the F2 RES control-input page, and use the other
parameters to assign various controls to alter it. If a boost is applied at frequencies where there
-80
-60
-40
-20
0
20
10 100 1000 10000 100000
Frequency in Hertz
Amplitude in dB
Cutoff frequency at C 5;
resonance from -12 to 24 dB
in increments of 6 dB
24
-12

16-10
DSP Functions
The DSP Functions
are high-amplitude partials, the signal may clip. The Pad parameter on the F1 FRQ page will
reduce the clipping, but thereÕs no harm in keeping it if you like the sound.
EditProg:F2|RES(2P|LOPASS)|<>Layer:1/1||
Adjust:0.0dB||||||||Src1||:OFF||||||||||
||||||||||||||||||||Depth|:|0.0dB|||||||
||||||||||||||||||||Src2||:OFF||||||||||
KeyTrk:|0.00dB/key||DptCtl:OFF||||||||||
VelTrk:|0.0dB|||||||MinDpt:|0.0dB|||||||
||||||||||||||||||||MaxDpt:|0.0dB|||||||
<more||F1|FRQ|F2|RES|F3|||||F4|AMP|more>
Two-pole Lowpass Filter, -6 dB Resonance (LOPAS2)
Using this Þlter is the same as using two one-pole lowpass Þlters in successive algorithm blocks.
Since its resonance is Þxed at -6 dB, itÕs also the same as using 2POLE LOWPASS with the
resonance set to -6 dB. YouÕd use this Þlter when you want a 12 dB per octave rolloff but donÕt
need to set a resonance level. This would leave you free to use another DSP function in the
algorithm, since itÕs a one-stage function.
Parameter Range of Values
Adjust -12 to +48 dB
Key Tracking ± 2.00 dB per key
Velocity Tracking -30 to +60 dB
Source 1 Control Source list
Source 1 Depth -30 to +60 dB
Source 2 Control Source list
Source 2 Depth Control Control Source list
Minimum Depth, Source 2 -30 to +60 dB
Maximum Depth, Source 2 -30 to +60 dB
-80
-70
-60
-50
-40
-30
-20
-10
0
10 100 1000 10000 100000
Frequency in Hertz
Resonance = 0 dB;
cutoff frequency
from C 4 to C 10
Amplitude in dB
C 4 C 10

DSP Functions
The DSP Functions
16-11
Two-pole Lowpass Filter, +12 dB Resonance (LP2RES)
This is similar to LOPAS2; the only difference is that its resonance is Þxed at +12 dB.
Four-pole Lowpass Filter with Separation (4POLE LOPASS W/ SEP)
-80
-70
-60
-50
-40
-30
-20
-10
0
10
20
10 100 1000 10000 100000
Frequency in Hertz
Amplitude in dB
Resonance = 12 dB
cutoff frequency
from C 4 to C 10
C 5 C10
-80
-70
-60
-50
-40
-30
-20
-10
0
10 100 1000 10000 100000
Frequency in Hertz
Resonance = 12 dB;
separation = 0;
cutoff frequency from C 5 to C 10
Amplitude in dB
C 5 C 10

16-12
DSP Functions
The DSP Functions
This combines 2POLE LOWPASS and LOPAS2 in one three-stage function. The parameters on
the F1 FRQ control-input page affect the cutoff frequencies of both Þlters. The parameters on the
F2 RES page affect the resonance of 2POLE LOWPASS. The parameters on the F3 SEP page shift
the cutoff frequency of LOPAS2, creating a separation between the cutoff frequencies of the two
-80
-70
-60
-50
-40
-30
-20
-10
0
10
20
10 100 1000 10000 100000
Frequency in Hertz
Amplitude in dB
Cutoff frequency = C 5;
separation = 0;
resonance from -12 to 24 dB
24
-12
4-Pole Lowpass Filter:
Resonance
-80
-70
-60
-50
-40
-30
-20
-10
0
10
20
10 100 1000 10000 100000
Frequency in Hertz
Amplitude in dB
Cutoff frequency = C 7;
resonance = 12 dB;
separation from -2 to +2
-2 +2
4-Pole Lowpass Filter:
Separation in Octaves

DSP Functions
The DSP Functions
16-13
Þlters. Positive values raise the cutoff frequency of LOPAS2, while negative values lower it. If no
separation is applied, thereÕs a 24 dB per octave rolloff above the cutoff frequency.
EditProg:F3|SEP(4P|LOPASS)|<>Layer:1/1||
Adjust:0ct||||||||||Src1||:OFF||||||||||
Fine||:0ct||||||||||Depth|:0ct||||||||||
||||||||||||||||||||Src2||:OFF||||||||||
KeyTrk:0ct/key||||||DptCtl:OFF||||||||||
VelTrk:0ct||||||||||MinDpt:0ct||||||||||
||||||||||||||||||||MaxDpt:0ct||||||||||
<more||F1|FRQ|F2|RES|F3|SEP|F4|AMP|more>
Gated Lowpass Filter (LPGATE)
You may be familiar with gates as applied to effects like reverb, where the effect shuts off
abruptly after a speciÞed time. The gated lowpass Þlter produces a somewhat similar effect in
terms of the soundÕs amplitude. The ÞlterÕs cutoff frequency is controlled by the AMPENV.
When the AMPENV is at 100%, the cutoff frequency is high, so most of the partials are heard.
When the AMPENV decays or releases to 0%, the cutoff frequency is low, so only the lowest
partials are heard. YouÕll hear the distinct effect of the Þlter closing as the amplitude envelope
releases.
Parameter Range of Values
Coarse Adjust ± 10800 cents
Fine Adjust ± 100 cents
Key Tracking ± 250 cents per key
Velocity Tracking ± 10800 cents
Source 1 Control Source list
Source 1 Depth ± 10800 cents
Source 2 Control Source list
Source 2 Depth Control Control Source list
Minimum Depth, Source 2 ± 10800 cents
Maximum Depth, Source 2 ± 10800 cents

16-14
DSP Functions
The DSP Functions
One-pole Highpass Filter (HIPASS)
High-frequency partials pass through this Þlter unaffected. At the cutoff frequency, the signal is
attenuated 3 dB. ThereÕs a roll-off of 6 dB per octave below the cutoff frequency. The resonance is
Þxed at -3dB. When the cutoff frequency is well above the highest-frequency partials of a sound,
raising the cutoff further will not affect the timbre of the sound, but will merely attenuate it
further.
The Coarse Adjust parameter sets the cutoff frequency in terms of a key name. The remaining
parameters (except Pad) alter the cutoff frequency in increments of cents. Positive key tracking
values raise the cutoff frequency for high notes and lower it for low notes. More speciÞcally, a
value of 100 cents per key on this page, when Þltering a constant waveform like a sawtooth,
would result in waveforms of exactly the same shape for all pitches of the waveform. The cutoff
frequency moves in sync with the frequencies of the waveformÕs partials as different pitches are
generated. Negative key tracking values will steepen the rolloff of highpass Þlters below the
cutoff. The Pad parameter, as always, attenuates the signal at the input to the function. These
parameters affect all the highpass Þlters similarly.
-50
-45
-40
-35
-30
-25
-20
-15
-10
-5
0
10 100 1000 10000 100000
Frequency in Hertz
Amplitude in dB
Cutoff frequency
from C 2 to C 7
C 2 C 7

DSP Functions
The DSP Functions
16-15
Two-pole Highpass Filter (HIPAS2)
This is very similar to HIPASS. The primary difference is in the steepness of the rolloff at the
cutoff frequency. Below the cutoff frequency, the rolloff is similar to that of HIPASS, except that
thereÕs a one-octave shiftÑthat is, HIPASS with a cutoff frequency of C 3 will sound nearly the
same as HIPAS2 with a cutoff of C 4. In other words, HIPASS gives you greater attenuation of
low frequencies when set to the same cutoff frequency as HIPAS2.
Four-pole Highpass Filter with Separation (4POLE HIPASS W/ SEP)
-50
-40
-30
-20
-10
0
10
20
10 100 1000 10000 100000
Frequency in Hertz
Amplitude in dB
Resonance = 0 dB;
cutoff frequency
from C 2 to C 7
C 2 C 7
-30
-25
-20
-15
-10
-5
0
5
10
10 100 1000 10000 100000
Frequency in Hertz
Amplitude in dB
Resonance = 0;
separation = 0;
cutoff frequency
from C 2 to C 6
C 2 C 6

16-16
DSP Functions
The DSP Functions
This combines two of the 2POLE HIPASS Þlters into one three-stage function. It has a rolloff of 6
dB per octave below the cutoff frequency. The parameters on the F1 FRQ control-input page
affect the cutoff frequencies of both Þlters. The parameters on the F2 RES page affect the
resonances of the Þrst Þlter. There will always be a slight extra boost of partials at the cutoff
frequency, even at low resonance settings. The parameters on the F3 SEP page shift the cutoff
frequency of the second 2POLE HIPASS, creating a separation between the cutoff frequencies of
the two Þlters. Positive values raise the cutoff frequency of the second 2POLE HIPASS, while
negative values lower it. If no separation is applied, thereÕs a 24 dB per octave rolloff below the
cutoff frequency. A variety of responses can be produced by adjusting the resonance and
separation settings.
-25
-20
-15
-10
-5
0
5
10
15
20
10 100 1000 10000 100000
Frequency in Hertz
Amplitude in dB
Cutoff frequency = C 5;
separation = 0;
resonance from -12 to 24 dB
24
-12
4-Pole Highpass Filter:
Resonance
-30
-25
-20
-15
-10
-5
0
5
10
10 100 1000 10000 100000
Frequency in Hertz
Amplitude in dB
Cutoff frequency = C 5;
resonance = 6 dB;
separation from -2 to +2
-2
+2
4-Pole Highpass Filter:
Resonance

DSP Functions
The DSP Functions
16-17
One-pole Allpass Filter (ALPASS)
Allpass Þlters do not affect a soundÕs frequency response (the amplitude of partials at various
frequencies), but change the phase of each partial depending on its proximity to the center
frequency. The phase shift is -90 degrees for partials at the center frequency. It rises toward 0
degrees for partials at frequencies below the center, and falls toward -180 degrees for partials at
frequencies above the center. With low-frequency waveforms, youÕll be able to hear this phase
shift. As a rule, however, the ear is not sensitive to phase shifts unless theyÕre changing, so youÕll
usually want to use Source 1 or 2, and assign an LFO to sweep the center frequency up and
down.
Periodic phase shifts like those induced by an LFO on the center frequency will cause a vibrato-
like variation in the pitch of a sine wave input. This vibrato effect will be less regular for more
complex partials. The greater the depth setting of the control source using the LFO, the greater
the vibrato effect.
The amount of vibrato effect also depends on the speed and amount of the phase shift. Try
adjusting the rate of the LFO controlling the center frequency. Another way to increase the
amount of phase shift is to use the two-pole allpass Þlter, or to use the one-pole allpass Þlter in
more than one algorithm block.
-180
-160
-140
-120
-100
-80
-60
-40
-20
0
10 100 1000 10000 100000
Frequency in Hertz
Phase in degrees
Cutoff frequency
from C 4 to C 10
C 4 C 10

16-18
DSP Functions
The DSP Functions
Two-pole Allpass Filter (2POLE ALLPASS)
Using 2POLE ALLPASS is very similar to using ALPASS, with two differences. First, the phase
shift is -180 degrees for partials at the center frequency, approaching 0 degrees for partials at low
frequencies, and approaching -360 degrees for partials at high frequencies.
Second, since this is a two-stage function, thereÕs an additional control-input page (F2 WID)
which controls the Þlter width. The parameters on this page affect the frequency range,
measured in octaves, where most of the phase shifting occurs. Small values cause a drop from 0
to -360 in the phase shift to occur near the center frequency, while large values spread the drop
in the phase shift over a broader frequency range. Small values tend to affect just a few partials,
leaving others mostly untouched. The affected partials seem to become detached from the
others, creating the illusion of an additional sound source.
-360
-340
-320
-300
-280
-260
-240
-220
-200
-180
-160
-140
-120
-100
-80
-60
-40
-20
0
10 100 1000 10000 100000
Frequency in Hertz
Phase in degrees
Width = 2 octaves;
Cutoff frequency
from C 4 to C 10
C 4 C 10
-360
-340
-320
-300
-280
-260
-240
-220
-200
-180
-160
-140
-120
-100
-80
-60
-40
-20
0
10 100 1000 10000 100000
Frequency in Hertz
Cutoff frequency = C 6;
width from .1 to 4 octaves
4.1
Phase in degrees

DSP Functions
The DSP Functions
16-19
If you leave the center frequency constant and assign an LFO to vary the width, partials with
frequencies above the center will shift their pitches in the opposite direction of partials below
the center frequency.
EditProg:F2|WID(2P|ALPASS)|<>Layer:1/1||
Adjust:0.010oct|||||Src1||:OFF||||||||||
||||||||||||||||||||Depth|:0.00oct||||||
||||||||||||||||||||Src2||:OFF||||||||||
KeyTrk:0.000oct/key|DptCtl:OFF||||||||||
VelTrk:0.00oct||||||MinDpt:0.00oct||||||
||||||||||||||||||||MaxDpt:0.00oct||||||
<more||F1|FRQ|F2|WID|F3|AMP|F4|AMP|more>
Two-pole Notch Filter (NOTCH FILTER)
Parameter Range of Values
Adjust 0.010 to 5.000 octaves
Key Tracking ± .200 octaves per key
Velocity Tracking ± 5.00 octaves
Source 1 Control Source list
Source 1 Depth ± 5.00 octaves
Source 2 Control Source list
Source 2 Depth Control Control Source list
Minimum Depth, Source 2 ± 5.00 octaves
Maximum Depth, Source 2 ± 5.00 octaves
-70
-60
-50
-40
-30
-20
-10
0
10 100 1000 10000 100000
Frequency in Hertz
Amplitude in dB
Width = 2 octaves;
Center frequency
at C 4, C 7, C 10
C 4 C 10

16-20
DSP Functions
The DSP Functions
The two-pole notch Þlter has two control-input pages, one for center frequency, one for width.
Partials with frequencies above or below the notch will be unaffected. Within the notch, partials
will be attenuated according to the width of the notch. The width is deÞned in terms of the
number of octaves between the points on the signalÕs attenuation curve where the attenuation is
3 dB (see the explanation of F2 WID for the PARAMETRIC EQ functionÑpage 16-26). For
example, if the width is set at four octaves, then the attenuation will be 3 dB at two octaves in
either direction from the center frequency. ThereÕs no attenuation of partials at more than two
octaves in either direction from the center frequency.
Two-pole Notch Filter, Fixed Width (NOTCH2)
The only functional difference between NOTCH2 and NOTCH FILTER is that the width of
NOTCH2 is Þxed at 2.2 octaves. This gives you a one-stage notch Þlter function.
-80
-70
-60
-50
-40
-30
-20
-10
0
10 100 1000 10000 100000
Frequency in Hertz
Amplitude in dB
Center frequency
width from .1 to 4 octaves
= C 6;
4
.1
2-Pole Notch Filter:
Width in octaves
-70
-60
-50
-40
-30
-20
-10
0
10 100 1000 10000 100000
Frequency in Hertz
Amplitude in dB
Center frequency
at C 4, C 7, C 10
C 4 C 10