Modification de Mutable Instruments Clouds

=[[Mutable_Instruments|Mutable Instruments]] Clouds=

Texture synthesizer

{| border="0"
| valign="top"|
[[Fichier:Clouds.jpg|none|200px|thumb|Clouds Front]]
| valign="top"|
[[Fichier:130642 2.jpg|none|200px|thumb|clouds]]
| valign="top"|
[[Fichier:Clouds back.jpg|none|200px|thumb|clouds back]]
|}

==Pitch-shifter/time-stretcher==

This engine is quite similar to the granular mode, except that it uses two overlapping grains synchronized with the most salient period of the sound. The grains are carefully spliced so that they mesh well with each other (a technique similar to the “deglitching” of early pitch-shifters). Modulating POSITION when recording is frozen will “scrub” through the audio buffer. Clouds’ uses classic time-domain methods which are not suitable for polyphonic or percussive material (unless this percussive material is breakbeats and you liked Akai samplers. Then: smile).

DENSITY creates a granular diffusion effect based on all-pass filters; and TEXTURE acts as a low-pass/high-pass filter. SIZE controls the size of the overlapping windows used for pitch-shifting and time-stretching – from an extremely grainy “drilling” sound to smooth bits of loops.

Sending a trigger on the TRIG input creates a clock-synchronized loop (when FREEZE is enabled) or stuttering effect – equivalent to applying a tempo-synchronized decaying envelope on the POSITION parameter.
Looping delay

==Looping delay==

The looping delay engine continuously plays back audio from the buffer without any kind of granularization. POSITION controls the distance between the playback head and the recording head (in other words, the delay time). Modulating POSITION will create effects similar to vinyl scratching or manual manipulation of tape. When FREEZE is activated, the content of the audio buffer is looped (stutter effect). POSITION controls the loop start and SIZE the loop duration.

DENSITY creates a granular diffusion effect based on all-pass filters; and TEXTURE acts as a low-pass/high-pass filter. SIZE controls the size of the overlapping windows used for pitch-shifting – fully clockwise for a smooth result that might smear transients, fully counterclockwise for a grainy, almost ring-modulated sound.

When FREEZE is enabled, sending a trigger on the TRIG input creates a clock-synchronized stuttering loop. Otherwise, the period of the trigger pulses sets the delay time – provided this delay is shorter than the recording buffer size.

==Spectral Mode==

Parameters in spectral mode:

POSITION = selects in which buffer the audio is poured (when FREEZE is not active), or from which buffer the audio is synthesised (when FREEZE is active).

Example:

Set POSITION to minimum value. FREEZE. You get a first texture. Set POSITION to maximum value. UNFREEZE. Wait for something else to happen in the incoming audio. FREEZE again. By moving POSITION you interpolate between the two textures which had been captured at the press of FREEZE. Depending on the quality settings there are 2 to 7 buffers laid out on the course of the POSITION knob. So it's a bit like morphing between FFT slices.

SIZE = change the coefficients of a polynomial that determines how frequencies are mapped between the analysis and synthesis buffers. It's like a 1-knob GRM Warp. Over the course of the knob it'll do spectral shifting, but also spectral reversal.

PITCH = pitch-shifting.

DENSITY determines how results from the analyzer are passed to the resynthesizer. Below 12 o'clock, there's some increasing probability that a given FFT bin won't get updated, causing a kind of partial freeze. After 12 o'clock, adjacent analysis frames are increasingly merged together (like a low-pass filter in the amplitude each frequency bin). At extreme settings, random phase modulation is applied to smooth things - giving you different flavours of spectral muddling/reverb.

TEXTURE does two things: below 12 o'clock, it increasingly quantize the amplitudes of the spectral components, like a very low-bitrate audio file (a long time ago I loved making super harsh noise textures by loading text files as raw audio files in an audio editor... then encoding as mp3 or real audio with super low bitrate to make it sound like some underwater brian eno). After 12 o'clock, it increasingly weakens the strongest partials and amplifies the weakest ones. This has the effect of making the spectrum more noise-like.

==CheatSheets==

===Clouds CheatSheet V3 ===
by Titanium

[[Fichier:Mutable Instruments-Clouds Cheat Sheet v3.pdf]]

===Clouds Cheatsheet===
by ???

all the modes + parasites firmware modes

[[Fichier:CloudsCheatsheet.pdf | CloudsCheatsheet.pdf ]]

===Clouds Illustrated manual===
by ???

[[Fichier:Clouds_Illustrated_Manual.pdf | Clouds_Illustrated_Manual.pdf  ]]

===Supercrysalis_memo_for_MI_Clouds===
by mercutio

[[Fichier:Supercrysalis_memo_for_MI_Clouds.pdf | Supercrysalis_memo_for_MI_Clouds.pdf  ]]

==Olivier's notes==

As for Clouds, I don't even know where to start! As a module designer, it's very frustrating to see that it has become popular for all the wrong reasons. It seems to me that for a large number of people the trajectory is: "ohh I feed audio into it and it makes ambient-ish muddy sounds on the out. (later) I'm tired of the muddy sounds, let's install the alternative firmware and see if it can do more". The key to use this module is to feed random source into its CV inputs to randomize the grain parameters (otherwise all grains are the same and wow, you're stacking 40 delayed versions of the same audio signal... no wonder it's muddy). Any "granularizer" plug-in will have randomization built-in, but Clouds doesn't have any randomization built-in because hey, it's a module, you're supposed to make it happen with external modulations... Here's an example of interesting use of the module - creating chords from any oscillator. https://www.youtube.com/watch?v=gSdmQ08JUGM. Another completely different sort of game is to have gate/trigger patterns in the FREEZE and TRIG inputs for stuttering/glitch effects. Yet how many times I've seen Clouds without any other connections than IN L, OUT L and OUT R...

There's a lot of negative influence from the "VST culture" - people considering that one module = one virtual instrument or one module = one effect. I'm partly responsible because I tend to add to each of my modules the extra ingredient (an internal modulation, an internal VCA, an internal exciter...) that allows them to be played on their own in a small system. But they acquire all they value when they play a role within a larger patch...

=liens=

playlist sur youtube : https://www.youtube.com/embed/cy34cBwM-Uo?list=PL54NASqAHmsRHRedsHZtW2QLKNsgGpmhu

clouds/parasites reference guide : https://drive.google.com/file/d/0Bx2WGFW47ykbSXJRNEFwRHhBN1E/view

== Tago Mago Notes ==
Q i’m wanted to see what other mod63/
73
sequence
tuning
this
EG
then
so
position
variety
and
do
send
to
sounds
the
the
variation
since
way
immensely,
for
even
little
grains
about
me
complete
Clouds
folks have found most useful in
pairing with the clouds.
R
VCAs. For example, Veils.
Many times I’ve seen beginners getting
started with Eurorack, getting
a Clouds and all kind of stuff but
no VCA, because they think of the
VCA as the final stage of a subtractive
synth, what you plug your VCF
in. Terrible mistake!
I apologize to the guy to which I
originally wrote the blurb that I’m
reposting here, for breaking the privacy
of an email conversation, but it’s
something I found myself explaining
dozens of times:
Let’s say you’re processing a sound
with Clouds.
Now you want each grain to have a
different transposition amount, between
-1 octave and +1 octave. To do
that, you send a random signal (eg:
white noise) going between -1V and
+1V to Clouds’ V/O input. So all the
grains are at a random pitch, producing
a very chaotic swarm of sounds.
Now you want the signal to progressively
reach a state where all the grains
are almost in unison (a bit like the
THX sound). To do that, you’ll need
a VCA, controlled by a slow envelope
or by a ramp down LFO, to reduce the
amplitude of the noise towards 0.
More generally, every time you play
with an attenuator / attenuverter on
a module and say to yourself “it would
be cool if I could use an LFO to do this
automatically”, you’ll need a VCA
controlled by this LFO to animate the
modulation amount.
This is particularly important with
Clouds which does not have attenuverters
on its CV inputs. If so far, the
only CVs you’ve thrown at Clouds
were full scale signals, you’ll suddenly
discover a new world of sounds by
sending low amplitude signals - to
add subtle randomization and variety
to the grains rather than getting
them to play through the full range
of the transposition, size or position
parameters.
Of course, an attenuator or attenuverter
will also allow you to get there
and explore subtle modulations, but
with VCAs, you can now automate
the movement from intense to subtle.
-------------------
Here’s what I do to make a nice subtle
drone (which I do a lot):
* Use one LFO to modulate slowly
(10 seconds or so?) through the
buffer, input to position. Maybe use
Maths channel 1 in cycle mode for
this. Set Clouds position knob to 0 if
the LFO output voltage is only positive,
attenuate if it’s above 5V or so.
* Set Maths channel 4 cycling also
quite slow, tweak rise and fall so that
EOC LED goes off very shortly (<0.5
out of 5 seconds or so). Connect
channel 4 output to Clouds freeze
input.
* Size to 75% (maybe modulated
with attenuated LFO).
* Texture to 75% (maybe modulated
with attenuated LFO).
* Blend + Stereo to 100%.
* Set pitch, feedback and reverb to
taste.
* Put any audio into the input
(maybe your favorite MP3 song),
make sure gain is not too high, keep
the 4th LED off. Don’t stop the audio
feed.
What this will do: grains out of different
positions of the buffer will be
heard back and forth, now and then
new audio will be added to the buffer.
----------------------
Clouds Parasite
Clouds Parasite is an alternative
firmware for Clouds. It adds the following
main features to the module:
The Oliverb, a new, full-featured
creative reverb mode based on the
reverb of the original firmware (itself
based on a description of the Lexicon
480L), and inspired in design by the
Erbe Verb, with CV control over all
parameters. It is largely modeless,
probably unrealistic and goes far beyond
your traditional reverb;
Resonestor, a polyphonic resonator
processor with built-in support
for your beloved Karplus-Strong
synthesis. Plucked or fed with your
favorite sound, it spits out strange
guitar sounds, lush chords, evolving
drones, and more.
More envelopes for your grains: the
Texture knob allows to select ramp
down or ramp up-shaped grains.
Square-shaped grains have steeper
edges for extra glitch;
Availability of much smaller grains
compared to the official firmware,
and asymmetrical grain envelopes
(saws, ramps).
Enhanced looping delay and time
stretcher modes: tempo syncing with
clock multiplier/divisor, support for
stereo delay (ping-pong delay) and
(crude) open feedback loop, much
improved sound quality, faster delay
time change.
---------
The procedure to access the unofficial
modes was a bit convoluted
(hold button 1 and press button 2) -
and could be easily confused with the
calibration procedure (hold button 2
and press button 1). The units which
have been shipped to dealers have
an additional “secret handshake”
for the extra modes which is easier
to remember: just hold the button 1
(BLEND MODE) for five seconds.
--------------
Parameters in spectral mode:
POSITION = selects in which buffer
the audio is poured (when FREEZE is
not active), or from which buffer the
audio is synthesised (when FREEZE
is active).
Example:
Set POSITION to minimum value.
FREEZE. You get a first texture. Set
POSITION to maximum value. UNFREEZE.
Wait for something else
to happen in the incoming audio.
FREEZE again. By moving POSITION
you interpolate between the
two textures which had been captured
at the press of FREEZE. Depending
on the quality settings there
are 2 to 7 buffers laid out on the
course of the POSITION knob. So
it’s a bit like morphing between FFT
slices.
SIZE = change the coefficients of a
polynomial that determines how fre64/
73 quencies are mapped between the
analysis and synthesis buffers. It’s
like a 1-knob GRM Warp. Over the
course of the knob it’ll do spectral
shifting, but also spectral reversal.
PITCH = pitch-shifting.
DENSITY determines how results
from the analyzer are passed to the
resynthesizer. Below 12 o’clock,
there’s some increasing probability
that a given FFT bin won’t get updated,
causing a kind of partial freeze.
After 12 o’clock, adjacent analysis
frames are increasingly merged together
(like a low-pass filter in the
amplitude each frequency bin). At
extreme settings, random phase
modulation is applied to smooth
things - giving you different flavours
of spectral muddling/reverb.
TEXTURE does two things: below
12 o’clock, it increasingly quantize
the amplitudes of the spectral components,
like a very low-bitrate audio
file (a long time ago I loved making
super harsh noise textures by loading
text files as raw audio files in an audio
editor... then encoding as mp3 or real
audio with super low bitrate to make
it sound like some underwater brian
eno). After 12 o’clock, it increasingly
weakens the strongest partials and
amplifies the weakest ones. This has
the effect of making the spectrum
more noise-like.
-----------------------
Using Clouds as an external effects
send on a Kontakt piano (The Giant)
was the starting point. Clouds
is in regular granular mode, with the
pitch knob almost all the way CCW,
and Texture at about 1 o’clock, and
Size at about 3 o’clock. There’s some
reverb and delay feedback, just to
glue things together a bit, but nothing
over the top.
To make things a bit more dynamic,
I patched a pink noise source into
Density and then took the same noise
into a sample & hold, and trigged the
sample from the note trigger from
the keyboard driving Kontakt. The
S&H output drove Position. This was
all done in Oscillot in Ableton Live,
rather than out in the modular rig - I
just fed the noise sources out through
a MOTU Ultralite.
This gave me a way of getting a different
“feel” to the grains for each
note, rather than it being a kind of
static, delayed tone. I was looking for
texture, rather than tones - which I
guess is the point of a “Texture Synthesizer”.
-------------------
Q quick question re clouds, I
haven’t
even used mine yet and I’ve had it in
my rack a month or more, can someone
tell me can I send it a trigger and
it trigger the entire sample that’s in
the buffer..? how do you set this up
if so..?
RS
o: you have audio in the buffer, because
when you pump audio into
it, there’s always audio in the buffer
(when I say ‘buffer’ we mean the
“live” RAM of Clouds - not the saved
memories). If you FREEZE this audio,
no new audio is put into the
buffer - the buffer is literally held as it
is. (You’ll need to freeze it if you want
the sample to play back, rather than
be overwritten by new audio).
Then: if you set DENSITY to 12
o’clock, so that no grains are sown,
when you put a trigger in on the
TRIGGER input, a single grain will
be sown. If you make the SIZE of that
grain big enough, and put the POSITION
to the very beginning, you’d
have something that played back a
sample.
Really, though, it’s not as good at
playing back the contents of the
buffer on demand precisely as it is at
playing them back vaguely/interestingly/
smeared out. So for instance:
you could also make the grains small
and fire a burst of pulses using a burst
generator/function generator. Or fire
a square wave LFO at it, opening the
LFO on demand with a VCA. Etc, etc.
--------------
1) If you’re going to modulate any of
the CV inputs, make sure you have
some attentuation happening in between.
Extreme CV’s will cause the
settings to slam again each of the far
ends of their reach when what you
really want is to create more subtle
movements (this applies especially to
Density and Size).
2) Be mindful of pitch settings. Depending
on your audio source, it can
get a little sloppy sounding if you’re
just keeping a static blend of the
source material and the grains that
Clouds is producing. Since the pitch
knob is quantized, it’s easy to find
pleasing little “plateaus”.
3) Use feedback sparingly. I almost
always have it set low or completely
off as I find it often sets the grains off
into a more unpredictable trajectory
(if that makes any sense).
4) Texture is your friend. The two
far extremes of this setting are exactly
that...extreme. Fully CW and
the grains start to stretch like taffy
(you may actually want this for pads,
etc), fully CCW and the grains can
be pretty harsh. Again, I find that setting
Texture to just before the really
smeary to be the most pleasing (to
my ears, at least).
5) If you’re looking for washed
out, though, the reverb setting will
definitely help. Don’t go too deep,
though. A little goes a long way.
-------------------
I really like sending audio rate modulation
to Position in granular mode.
My sweet spots tend to be Position
- anywhere, Size - between 1 and 5
o’clock, Pitch - noon, Gain - wherever
needed, Density - between 9 and
11:45 o’clock (or 12:15 and 3), Texture
- between 9 and noon, Wet/Dry
- 100% wet, Stereo Spread - as needed
(stereo is a must with Clouds), Feedback
- as needed but usually none or
just a bit, Reverb - as needed. I’ll often
mult out the incoming signal to have
a dry signal, or to send it to another
effect/filter or whatever. Works better
for me than using the wet/dry knob,
which I do love the new behavior.
All of the different modes and Parasite
firmware are worth exploring.
My favorite mode (outside the
granular stuff) might be the looping
delay mode. I almost always
drop the quality down to one of the
65/73
you
can
you’re
the
that
pitch
find
completely
trajectory
and
pads,
setting
washed
deep,
modulation
mode.
Position
wherever
and
Texture
often
another
Parasite
exploring.
looping
8-bit modes. At 100% wet and Position
(delay time) and Feedback at
0, you have this bit reduction, Lo-fi
mode going on. There’s a lovely flavor
of “tape hiss”, the filter (Texture)
and when you crank the gain you
get some awesome distortion (see r
beny track called Wilt for an example
of this). With longer delay times,
the pitch starts to warble in a pleasing
way. There are some sweet spots
with Feedback, right before it starts
to feedback into noise, where it will
repeat for a very long time, but slowly
decay the repeats into a blur. Really
fun for some Basinski-like tape
degradation.
--------------
As the title suggests, this is actually
three Clouds used in series. If
I remember correctly, it’s Parasites
Resonestor mode, into standard
Clouds Granular mode, into Parasites
Oliverb mode.
The one that’s handling the granular
functions is probably set to what
I like to refer to as my “zero point”.
Basically, I’ll have it in the second
audio quality setting, have position
set back to ~2:00, keep grain Size
fully CW, and texture just right before
the last algorithm (slight smearing
of the grains, but not to “gauzy”).
From here, I’ll often set the pitch to
a higher quantized interval/octave
of the incoming melody/sequence/
whatever sound to give it a more
“harmonic” feel. CV’ing the Density
will then give it movement. I’ll often
used an attenuated (very important)
slow random sine from Peaks for that
purpose. Also, Blend is generally set
right at 12:00, stereo spread fully CW,
feedback at zero, reverb set to just beyond
12:00.
Though, keep in mind, the Clouds
that’s set to Oliverb is doing a lot
of the heavy lifting in the example
above. That’s creating more of the
shimmery space that you hear wandering
around. I no longer have the
third Clouds, but I often use Granular->
Oliverb chain in my patches.
So here are the takeaways:
1) If you’re going to modulate any of
the CV inputs, make sure you have
some attentuation happening in between.
Extreme CV’s will cause the
settings to slam again each of the far
ends of their reach when what you
really want is to create more subtle
movements (this applies especially to
Density and Size).
2) Be mindful of pitch settings. Depending
on your audio source, it can
get a little sloppy sounding if you’re
just keeping a static blend of the
source material and the grains that
Clouds is producing. Since the pitch
knob is quantized, it’s easy to find
pleasing little “plateaus”.
3) Use feedback sparingly. I almost
always have it set low or completely
off as I find it often sets the grains off
into a more unpredictable trajectory
(if that makes any sense).
4) Texture is your friend. The two
far extremes of this setting are exactly
that...extreme. Fully CW and
the grains start to stretch like taffy
(you may actually want this for pads,
etc), fully CCW and the grains can
be pretty harsh. Again, I find that setting
Texture to just before the really
smeary to be the most pleasing (to
my ears, at least).
5) If you’re looking for washed
out, though, the reverb setting will
definitely help. Don’t go too deep,
though. A little goes a long way.
---------------------
The key to use this module is to
feed random source into its CV inputs
to randomize the grain parameters
(otherwise all grains are the
same and wow, you’re stacking 40
delayed versions of the same audio
signal... no wonder it’s muddy).
Any “granularizer” plug-in will have
randomization built-in, but Clouds
doesn’t have any randomization
built-in because hey, it’s a module,
you’re supposed to make it happen
with external modulations... Here’s
an example of interesting use of the
module - creating chords from any
oscillator. https://www.youtube.com/
watch?v=gSdmQ08JUGM. Another
completely different sort of game is
to have gate/trigger patterns in the
FREEZE and TRIG inputs for stuttering/
glitch effects. Yet how many
times I’ve seen Clouds without any
other connections than IN L, OUT L
and OUT R...
----------
- play some varied audio into the
buffer - an arpeggio is good because
you will know where you are in the
loop
- press freeze - this retains the last 4
seconds of what you played
- turn position fully counter clockwise
- this now means the play point
of clouds (imagine a tape head) is at
the end of the recording.
- Now put a slow sawtooth LFO into
the position socket. This will then
move the position from the start of
the buffer to the end at the speed of
the LFO.
- Put grain density a little bit left of
centre until you start to hear individual
grains come out. Every grain
will be from a different position in
the buffer because you are modulating
position, and because you are using
a sawtooth LFO, this will play the
buffer back in the same direction you
recorded it.
- So you will end up with a series of
fragments of the original recording
in sequence
- play with the size knob to make the
grains smaller or larger and you will
start to hear more of the original.
---------
QC
louds’ buffer is constantly being fed
by the incoming audio, using triggers
or gates to rhythmically freeze
the buffer continuously will build up
a “cloud” of audio that’s an accumulation
of the sounds that have been
fed into the buffer and smeared together.
This cloud becomes an evolving
foundation that incoming sounds
play with or counterpoint to and then
become as they are fed into the buffer
66/73 and smeared.
I was doing this with delay by using
feedback that would hold things long
enough to let them smear but eventually
let things pass - the in between
time is the space in which the structure
created itself.
Using Clouds, this worked well for
me by 1) feeding a song/audio with
different sounds (actually with the
delay it was an infinite delay of a filter
resonating, which was basically an
oscillator) 2) using triggers to freeze
the buffer at fairly quick intervals 3)
modulating position and size, along
with the blend of the dry/wet 4)
having enough feedback and reverb
to smear the sound but not create a
mud blob
Clouds makes this interesting because,
with the position and size (as
well as obviously texture and density)
randomly modulated, the sounds
that get frozen and smeared create a
variety of artifacts as they enter that
space.
R Smearing is only one extreme of the
granular synthesis, it occurs when
grains are either long, have a flatter
(longer envelope) and are many at
the some time with a similar content
and pitch. Typically composers like
Curtis Roads or Vaggione tries to
avoid smearing and either have
- Smalll grians
-Variable pitch
-Sharp envelopos
-MANY GRAINS
or
-long grains
-time stretching the underlying audio
- not many at one time
Naturally you can combine a large
slow grain with faster grains, the long
grain is then a backdrop. Since digital
grains of the slowed down long variety
tends to be harmonically complex
it often sounds smeared to have
several of these at the same time.
The idea behind granular synthesis
is to take a sample.. modulate speed
(and ther by pitch, one can also slow
a sample without altering pitch by
quickly looping, a loop a a few MS
the move the loop over the sample
(clouds don’t do this.. (or??)), apply
envelope over the sample, that creates
one grain. The character of the
grain varies depending on the underlying
sample, how that is treated
(slow, fast, freezed) and the shape of
the envelope. The cloud gets it character
from the individual grains and
the density of the grains..
So most composers that use granular
synthesis strives to avoid to much
smearing and Roads and Vaggione
composition are often crisp and very
dynamic.. and THAT is what i try
to achieve with clouds AND other
modules.. OK smear can also be cool
when you get a ghostly whisper or
dragon breath sound.. so to me the
great thing with Clouds is that you
can, if you want, avoid the smear..
and go for crisp and edgy..
--------------
A-118 colored noise --> Clouds in L
A-118 random --> Clouds size
Disting sine wave --> Clouds in R
Grids trigger 1 --> Clouds trigger
Grids trigger 2 --> Clouds 1V/oct
Grids trigger 3 --> Clouds texture
Grids accent 1 --> Clouds panning
blend
Key settings for Clouds:
Granular mode
Density at 12 o’clock
Dry/wet blend fully clockwise
Size between 7 and 9 o’clock
*********************

==videos==
https://www.youtube.com/watch?v=cy34cBwM-Uo&list=PL54NASqAHmsRHRedsHZtW2QLKNsgGpmhu


== Calibration ==

To calibrate the unit:

*Disconnect all CV inputs.
*Connect the note CV output of a well-calibrated keyboard interface or MIDI-CV converter to the V/OCT input.
*Press the Load/save button, and while you hold it down, press the Blend parameter/Audio quality button. The first 2 LEDs will blink in orange.
*Play a C2 note, or send a 1V voltage from your CV source.
*Press the Load/Save button. The four LEDs will blink in orange.
*Play a C4 note, or send a 3V voltage from your CV source.
*Press the Load/Save push-button.

Calibration is done!