Rev. 5 – Jun 2020 Page 20 of 91
external noise which may be used for fricative/sibilant
sounds.
SPECTRAL TRANSFER: A BRIEF HISTORY
In the modular synth domain, the most famous device
for performing spectral transfer was the legendary Model
296, designed by Donald Buchla in the seventies. It was
a 16 bands equalizer, which featured one input for the
odd bands and one for the even bands: this solution al-
lowed the artist to process two different signals at the
same time, but it also allowed to transfer the harmonic
content of one to the other via dedicated switches. It was
thus possible to analyze the odd bands and transfer their
amplitude to the even bands, and vice versa.
This design was clever because it took advantage of a
single array of 16 filters, but the resulting spectral transfer
was somehow “approximated” because the modulation
signal came from the analysis of the adjacent band, i.e.
another frequency area of the spectrum, albeit quite
close.
The FUMANA pushes this concept even further, but
still within the analog domain. An additional bank of 16
filters is added, specifically for analysis purposes, with a
different band-pass slope (see below §5).
When performing a spectral transfer, it is fundamental
to consider the harmonic content in both main and mod-
ulation audio signals. Remember that if there is no har-
monic to excite the content on the main signal, there will
not be that much outcoming audio, even if all the VCAs
are opened at their maximum level:
Poor harmonic content on the Main signal: if a pure sine
(one harmonic) or any signals with poor harmonic con-
tent is used, there are not many chances to hear good re-
sults exciting several bands. Supposing it is a sine wave
with a frequency of 105Hz (which lies then on band 3),
theoretically something happens only when band 3 is ex-
cited from the modulation signal.
Rich harmonic content on Main signal: this means more
chances to have several bands excited by the envelope
followers control voltages created by the modulator fil-
terbank.
Poor harmonic content on Modulator signal: allows cre-
ation of very selective envelope followers signals, which
translates in an extremely selective spectral transfer. If the
modulation signal changes pitch or has a very variable
content, it will for sure result in a more heterogeneous
spectral transfer.
Rich harmonic content on Modulator signal: allows a
very rich spectral transfer. Just keep in mind that if you
are using a square wave in the modulation with frequen-
cies varying from 100 to 200 Hz, probably you won’t hear
that much difference, due to the richness of harmonics of
the square wave.
PANEL OVERVIEW
A consistent color and graphic coding make the front
panel easy to understand at a glance, once properly un-
derstood.
The main graphic solution is the distinction between
the odd and even bands: since the two groups can work
independently, every input or output related to the even
bands is marked by a circle around the jack socket. (The
bands are numbered 1-16 from left to right: odd bands
are band 1, 3, 5, 7, 9, 11, 13, 15; even bands are 2, 4, 6,
8, 10, 12, 14, 16).
Another key coding is the distinction between the Main
filter array and the Mod filter array. Everything that re-
lates to the Main filter bank, which is the circuit directly
affecting the sound that is heard, is marked by the color
blue, let it be audio output, CV input or even the faders’
LED color.
On the other hand, everything that relates to the Mod
filter bank, which is the one that extracts the envelopes
from the modulating signal and modifies the harmonic
content of the main filter bank, is marked in grey, includ-
ing the individual envelope output LEDs (which flash
white when active).
Finally, the green and yellow colors mark the Global
Spectral Editing tools: the former relates to the Tilt con-
trol, the latter to the Scan controls.
2 AUDIO INPUTS
FUMANA has two pairs of inputs, which are called
Main In (blue), and Mod In (gray), short for “modulation
input”, plus a fifth input called Unvoiced. (A.1, A.3, A.5,
A.7, A.9).
Each pair is composed of inputs for the Odd and Even
bands, which are semi-normalled together. This means
that when only one patch cable is connected to one of the
two inputs, it automatically feeds the other. If you want
to use different sources for odd and even bands, simply
use two different cables. If instead, you want to feed only
the odd bands and not the even (or vice versa), simply
feed the odd input, and plug a dummy cable to the other
input (or vice versa).
Each of the four inputs has its own amplitude potenti-
ometer (A.2, A.4, A.6, A.8): different levels can thus be
set for odd or even filters within each pair of input. This
also means that if you want to emphasize even bands on
the main signal, you can simply add more gain to them,
and less to the odd ones.
The fifth input, named Unvoiced, is designed for adding
some depth to the fricative consonants that might be
missed when performing vocoding-like operations. It fea-
tures its gain control as well (A.10).
The red LEDs connected to the odd and even inputs
light up displaying the amplitude of the incoming audio
after the gain level.
To Next Page
Loading...