I'm really impressed with the granulator, good work !
A few suggestions about it :

- Granulator : add feedback in the granulator because the MXXX feedback module takes a lot of CPU and the granulator sounds great with feedback.
Maybe each voice could have it's own feedback control.

- Granulator : do you think it would be possible to modulate a point in the transformation graph ? X axis and Y axis modulation of one point of the curve. It does cool sounds when changing the transformation graph editor curve with the mouse in real time !

- Granulator : voice pitch -36 to +36 (cf soundtoys crystallizer) or add an octave slider just like in the harmonizer. I understand it might be difficult to fit more than -12/+12 in the sequencer panel, it would be great to have more though.
What about a pan on each voice for huge stereo effects (instead of using a utility module on each output to pan things).
I believe the ability to pan the voices and the feedback to be features of the granulator effect rather than options depending on other MXXX processors.

- Granulator : in the Advanced menu, "random pitch settings" is a bit confusing, at first i thought it was associated with "random pitch" or "random transformation". Maybe "Sequenced pitch settings" would be better?

- Granulator : when the sequencer mode is activated in the advanced panel, the pitch and pitch random sliders don't seem to do anything, maybe when the sequencer is activated, the pitch and pitch random sliders should be disactivated in the gui to reflect this.


The sequencer is really great.
As i understand, Transformation n°1 reverses the grain.
So i guess the horizontal axis of the transformation graph editor is the time period (= to grain size), and the vertical axis represents the start and end of the grain ?
I'll say it again, i think it's fantastic

I'm afraid it's not that simple. It will end up with more or less the same CPU load as with feedback. I'll check if it can be added just for convenience.



Hmmm, I was experimenting with that long time ago and it didn't end very well . But I'll check, it could be useful in some plugins.



With the pitch range - I'll check. I'm not sure if it will sound reasonable.
With pan and alike - the trouble is that number of possibilities is basically limitless, so there you could come up hundreds of other ideas, that's what you can easily do with other processors, and that's what's MXXX' purpose . I'll check the pan though.



But it is associated with random pitch . But it lets you do more than that - but I can see the modes are not documented . But well, let's rename it to Pitch sequencer .



Not really, they can mark the range for the "limited" modes. With "unlimited" modes they are useless, so that may be reasonable.



Thank you! And yes , horizontal is source position, vertical is target position (same thing with different words )

Thank you for the precision about the random pitch control.

With a little more time spent using the granulator and comparing it to other similar plugins, I'll say it's certainly a particular and interesting approach on the granular thing especially with the Transformation and sequencer.

"But", one thing i regret about it is the fact that "only one grain is payed at each moment" and if i understand correctly, it does not play the different grains one above the other at the same time.

A cool thing about the other granulators is that they can play the different grains at the same time (but i suspect each grain is played with a little delay relative to one another), so it does a really nice smoothed out sound. With feedback and a lot of grains (like 40 in KT granulator) they can do sounds very comparable to a synth pad sound.

This is not really possible at the moment with the granulator in MXXX.
I was wondering if there is a technical reason why the grains can't be played at the same time in the MXXX granulator ?

I was thinking maybe Since MXXX granulator has 4 voices, maybe voices 3 and 4 could be a little different and play all the grains at the same time ?

Okay i promise this is my last "feature request" until the final version of MXXX.

Well, the thing is, this is more like 2 things at once - I checked the Uhe's granulator and it creates also just one voice, simply because without pitch-shift it must work that way. But it could layer voices I guess... we'll see .

Ok I need some help here - I'm checking the KT Granulator and can't seem to find a way to make it produce more voices at once. How to do that?

Little bug....
When selecting a square in the grain sequencer I have to select the bottom right corner of the square or a square to the left and above the wanted square is selected.

Win7 home 32 bit
vst 2.4 plugin
ableton live v8.2.6 and 8.3

It might be wrong but from the way i understand the KT granulator by listening to it ; it's like 1 grain = 1 voice.

If you use like 20 grains, all the grains are played at the same time from start to finish (it's more obvious if you set the Duration -Dur.- to max). The grains sort of melt together and that's why it does this sort of jaw droping beautiful synth like textures especially when there's is a slightly different delay between left and right channels(the KT randomly sends the grains left and right from what i understand).

Grain 1
.Grain 2
..Grain 3
...Grain 4
....Grain 5
.....Grain 40
etc

The KT has control of the timing of each grain start, each grain duration, attack and release of each grain + triangle/rectangle envelope shape of each grain.
If the duration is not set to max, the KT granulator will act more like the melda granulator (only one grain active at the same time because the grain is not playing from start to finish).

Gra > Gra > Gra > Gra ...

In the melda granulator, if you use only one voice, it's seems the grains are not played simultaneously but successively one after the other ?

Grain 1 > Grain 2 > Grain 3... etc
That's why you can do sequenced melodies by pitching each grain to a specific pitch.

So the melda is better to do syn/co/pa/ted and sequenced sounds, and the Kt is better at doing smooooooooth textures (like when you loop/overdub one note over and over again to do a huge wall of sound).

Now that's what would be cool :

Grain 1.0 > Grain 2.0 > Grain 3.0
.Grain 1.1 > Grain 2.1 > Grain 3.1
..Grain 1.2 > Grain 2.2 > Grain 3.2
...
....Grain 1.40 > Grain 2.40 >

You could control the number of superposed grains (number of voices in the melda granulator) + the number of successive grains (graincount or steps in the sequencer). Apparently, to save CPU, the KT granulator automatically adjusts the number of grains active at the same time, up to 40 grains for superposed grains (high duration), only 1 for successive grains (low duration). It's quite something now that i think of it.

Again, maybe i'm wrong, it's just the impression i get from listening to the different granulators.


VariKusBrainZ > you're right i noticed that too at first, it's hard to select the correct square but i got adjusted to it and forgot about the problem, it's good you mentionned it.

not sure if i understood it correctly. but it seems i can't replicate it on mac osx. it's located in the grain sequencer under the tap "random pitch setting", right? clicking on a square always selects the meant square.


bests

The thing is I was unable to make KT granulator to play multiple voices at once. So I'm not really sure what is the max count for.
Anyway, yes, it could be possible to add multiple voices having the same parameters. The 4 voices was originally intended to be used that way.

Try this setting :


You will hear the 40 grains (or voices) playing at the same time.
It sounds somewhat similar to a 40 tap delay, but because it only plays chunks of the sound and because you can smooth out the attack of those grains it gets closer to a reverb.

If you set both delay sliders at the same position, you don't hear the effect because all 40 grains play at the same time ! They need to be played with a little delay between each of them of course.

From the KT manual :
Delay time: determines start of grain position (in s).
Most sliders are grouped in pairs: the first one specifies the minimum, the second one the maximum.

So basically in this setting, the delay time is sort of random between 0 and 3.125 (whatever that is in milliseconds).

Now if you really want something beautiful, set the pitch to +12, and by tweaking the shape and skew and feedback you can get smooth synth like sounds from a guitar or other instrument...

You can also try this for 10-15 active grains instead of 40.


>> Also it's very useful to have separate grainsize control for left and right channels, for bigger stereo sounds. The soundtoys has a "grain size offset" to do this and the argot lunar has a similar feature.



Basically yes if each voice of the melda granulator could be multiplied 40 times with the same parameters and some sort of delay control to set up the starting point of all those voices, it would become four KT granulators in one plugin + the transformation and sequencer.
Last edited by farfadetfarfelu on Sat May 26, 2012 9:25 am; edited 4 times in total

I think the automatic grain number in the KT granulator works this way :

If the grain size is superior to the grain duration, only one grain is needed, because the lenght of the grain is long enough and will need to be shortened/divided anyway.

If the grain size is smaller than the grain duration, then the plugin uses multiple grains playing at the same time to artificially create a longer grain from blending together the shorter grains with some delay between the grains...

Well, next version of the granular will have the multiple voices with the same parameters, so we'll see

You work really fast.

So i tried the updated granulator in 6.01b3.

The modulation of the transformation graph is stunning.
The additionnal voices copies do what is expected, bigger layered sounds. Cool !
The pan and random pan are perfect.

The delay control in each voice in now a little glitchy on my system when using a lot of voice "copies".
BUT it's a lot less glitchy when using the "minimal protection" option in the settings. I don't thing the analog style delay detuning/tape slowing down effect is really important in this delay control because you can get this with the transformation. There's probably a way to make it maybe less precise but also glitch-free ?

>> If you can add a random delay control to make more movement between the voice copies, it will be a very flexible granulator that can compete with most others out there including the GRM tools shuffler.

If it's not possible, well, you've already achieved something very nice with the particular transformation window, and i'm certainly very pleased with it!

Thanks!

The reason the delay is not working well now is that if you set say 20 voices and 50ms delay, then the first voice has 50ms, second 100ms... 20th voice has 1000ms, and then you change it... and you have 20 delays, where the last one is pretty big...




You mean like a separate delay control for each voice?

Ah i see.
I'll go on a limb and have a guess here (i have no idea if what i'm going to say is relevant but...) maybe the delay time could be fixed while the voices are playing, and only change "between" the grains, at the time interval when the grain is resampled ?



I mean just one "random delay" parameter (similar to random pan), so that each voice copy is always delayed by a random amount.

For example in random mode with a random interval from 10ms to 100ms:

> First grain sample
voice copy 1 will play 50ms after the original sound
voice copy 2 will play 70 ms after the original sound
voice copy 3 will play 10 ms after the original sound
...

> next grain sample
voice copy 1 will play 30 ms after the original sound
voice copy 2 will play 100 ms after the original sound
voice copy 3 will play 60 ms after the original sound
...

All grains are always randomly delayed as respect to the orignal grain.