~ICMC 2015 - Sept. 25 - Oct. 1, 2015 - CEMI, University of North Texas
under 15 milliseconds to maintain fusion with the original
sound [6]. To finish the basic design, first order low pass
filters are placed in each delay branch. The filter frequencies
(absorption) and the feedback gain (decay) are set uniformly
for all branches of the network so that the reverb is kept
balanced without any one delay time becoming dominant.
input
+ + + +
absorption-.
Svan vari vai var'
a a a a
ddd d
I G III
--
I I I I
~ -
decay --- --- x I I x I x 1 I x
_--- I ' I
- early reflections
Figure 1. Basic reverb structure of Erbe-Verb showing allpass filters (a), delay lines (d), low-pass filters, feedback
matrix and early reflection taps.
3. EXTENSIONS TO THE FDN.
3.1 Parameter Range and Modulation
The first extension of this basic reverb design comes from
simply giving the control parameters a large range, and attaching control voltage modulation to these parameters. For
instance, the delay time is widely variable, and can be
changed by signals as fast as 500Hz. The total delay variation ranges from as large as a football stadium to as small as
a closet. The internal delay times and all early reflection
times are continuously variable and scaled with the same
parameter. This modulation will create a mass of coordinated doppler shifts within the reverberated signal for a
dramatic change when changed slowly. When the reverb
delay time is modulated quickly, percussive effects can be
created, and when modulated at audio rate, FM effects can
be heard. Recently I saw a synthesist (Alessandro Cortini)
using the Erbe-Verb in just this way, creating cymbal-like
sounds with fast changes of the delay time. Other parameters can be modulated as well, adding doppler shift, tremulous sustain, etc.
3.2 Infinite Reverb
Creating long and infinite sustain is one of the primary goals
of the Erbe-Verb. The feedback gain (decay) is given a
maximum gain of 1.25, which can quickly put the reverb
network into infinite sustain. This feedback is controlled by
simple saturation functions (based on a 3rd-degree chebyshev polynomial), placed before the feedback matrix. The
harmonic generation of the saturation stages is balanced by
the lowpass filters. Consequently, because these stages
counteract each other, varying the decay and absorption
controls can create a large variety of sustained and infinite
reverb effects. The Erbe-Verb also generates a control voltage corresponding to the overall level of the reverb, which
can be used to negatively modulate the decay (like an audio
compressor with a feedback topology), and control the sustain without saturation.
3.3 Periodic time modulation
Modulation of the delay time and allpass filter time has been
used in many reverb designs to break up resonances from
the ringing of the allpass filters or to stop the pulsing that
comes from the combined overall delay time. Usually this
time modulation uses a sine oscillator or random function to
avoid the constant doppler shift which results from modulating by a constant slope waveform. The depth of modulation
is kept low to minimize chorusing effects. In my design, I
make available two types of modulation, with speed and
depth completely under control of the user. The first type is
a multiphase sine modulation section, with separate sine
modulation for each of the four delay lines. The frequency is
variable from 30 to.1 Hz., but can be synchronized to an
external clock for a wider range, or to connect the modulation to the tempo of a piece. The modulation depth varies
from 0 to 100% of the delay time.
3.4 Random time modulation
The Erbe-Verb also contains a random modulation scheme
which is inspired by Gordon Mumma's piece [7] "Stressed
Space Palindromes." In this piece, the room size and shape
is changing randomly and quickly from sound to sound. My
scheme uses four layer granular/raised cosine envelopes on
each delay line, with each grain selecting a random delay
time, constrained by the modulation depth. The modulation
speed is the grain creation speed. This type of modulation is
effective at breaking up resonances and pulsing, without
adding any chorusing or doppler shift. At extreme settings
(when the modulation depth is high) it imparts it's own
characteristic sound. At a slow modulation rate one can eas
- 263 -
