Updated - Mini now free!: FactorSynth & Factoid - M4L FSU/'sound unmixing' *essentials*

[cron]

[edit 3]FactorSynth 2 is coming soon and will be free for FactorSynth 1 owners.
--------------------------
[edit 2]The generously featured FactorSynth Mini is now free! Discussion begins on page 2.
--------------------------
[edit]New product released: Factoid time randomiser. See page 2 of this thread!
--------------------------
It's a long while since I've wanted to keep a discovery to myself, but it's not really fair on the developer and also conflicts with my wanting to shout about this thing from the rooftops.

In short, this thing is NUTS. It can rearrange and pull details from sounds like nothing else. Very, very RAM heavy and a little on the crashy side, but there's nothing quite like it and I'm getting fresh sounds I'm not sure I'd be able to create in any other way out of it. I've been using it for a month or so now and it already seems irreplaceable. I haven't even got over the honeymoon period of just wanting to play with it and surprising myself with what it can do yet.

https://www.youtube.com/watch?v=Vjbe_vJth38

[Dirtgrain]

Oh, that has my eye. Is it then a morphing device, similar to Morph 2 or Mmorph? Could you explain in some detail some of the things you have done with it?

From the website (https://www.jjburred.com/software/factorsynth/), it uses "use a data analysis algorithm called matrix factorization to decompose any audio clip into a set of temporal and spectral elements. By rearranging and modifying these elements you can do powerful transformations to your clips, such as removing notes or motifs, creating new ones, randomizing melodies or timbres, changing rhythmic patterns, remixing loops in real time, creating complex sound textures..."

[Puddi]

What the hell.

That looks amazing.

[whyterabbyt]

saw this a wee while back, forgot about it, so posting to remind me...

[cron]

Oh, that has my eye. Is it then a morphing device, similar to Morph 2 or Mmorph? Could you explain in some detail some of the things you have done with it?

It can do a spin on the morphing type sound - selecting each element in a diagonal line in the cross-synthesis half of the plug typically produces an unusual spin on the 'spectral vocoder' type sound you hear in morphing plugs. It's much more fun to rearrange the elements of input sounds using the randomiser in the left half of the plug though. The results can be anything from an unpredictable time/frequency rearrangement of the original sound through to something that sounds like a bizarre filtering/compression effect when the process 'fails'.

Being able to render out the individual elements with one click is really nice too. It allows you to lay them out in your DAW and process 'internal elements' of sounds, like a turbocharged version of Eventide's Physion or the more unusual split types in Melda's multiband plugs.

I've just moved computers so I haven't tried this yet, but I'm also planning to decompose sounds and load the elements into multisource capable granular synths (e.g. Kaivo, Hourglass), with the aim of producing something akin to a 'timbral arpeggiation' effect. There's probably a whole new class of interesting effects to be found with granular synths or standard samplers once you load the decomposed elements into them and play around with selective triggering or crossfading etc. Effects that come from being able to treat a single sound as a multisampled instrument, essentially.

[CrystalWizard]

It can do a spin on the morphing type sound...

...I've just moved computers so I haven't tried this yet, but I'm also planning to decompose sounds and load the elements into multisource capable granular synths (e.g. Kaivo, Hourglass), with the aim of producing something akin to a 'timbral arpeggiation' effect. There's probably a whole new class of interesting effects to be found with granular synths or standard samplers once you load the decomposed elements into them and play around with selective triggering or crossfading etc. Effects that come from being able to treat a single sound as a multisampled instrument, essentially.

Wow, that seems like a lot of sound design fun!

[Dirtgrain]

Thank you for the information.

[cron]

Just wanted to give this stunning software a little bump. Had a brief chat with the developer and version 2 is currently in development. Hugely excited to see what that will bring to the table.

[pottering]

Just look a this:

https://www.youtube.com/watch?v=KgpEGUoESA0

As cron said, it is just crazy.

[Edit:] I'm referring mostly to the last 20 seconds, with the simple but useful mix knobs, and that craaaaazy thing of (re-)drawing the temporal shape of just the attack.

[psynical]

The previous video is even crazier:
https://youtu.be/kd024Ssy0EY

How would do that without Factorsynth? Is that like an envelope follower with a gate triggering the reverb?

[carrieres]

crazy synth !

[cron]

I'm referring mostly to the last 20 seconds, with the simple but useful mix knobs, and that craaaaazy thing of (re-)drawing the temporal shape of just the attack.

It gets better - the vertical column lets you redraw the frequency content! So not only can you reshape the attack in time, you can reshape the frequencies your attack contains, or both at once. The frequency shapes sound quite 'static' when you redraw the temporal envelope, yet everything gels together beautifully and reproduces the original sound when left default. It's amazing how complex signals can be broken down into such simple components.

How would you [apply selective reverb] without Factorsynth? Is that like an envelope follower with a gate triggering the reverb?

In terms of doing it automatically (i.e. not just automating a send), Unfiltered Audio's G8 has 'reject outputs' that let you losslessly (i.e. with perfect quality) split the gated and ungated signals across 2 channels. So you could add reverb (or any effect) to only the loud parts or only the quiet parts. Obviously it doesn't 'unmix' components like FactorSynth does though. Depending on how FactorSynth's algorithms feel about the signal, you could potentially have two notes sounding simultaneously and only apply reverb to one of them. The closest I can think of is probably using one of the Melda multiband reverbs with the bandsplitter in tonal/transient mode. At least you're getting some sort of 'spectral' split there, but it's nowhere near as granular as FactorSynth can get where you can end up with multiple tonal, transient, noise etc components.

You do have to bear in mind that FactorSynth isn't for track-length processing though. It is very RAM hungry, so while you could conceivably load an entire track into it, you're unlikely to be doing so on today's machines, particularly when you're using large numbers of components.

For me, I'm sure it has all these amazing practical uses, but it's the out-there sound design possibilities that really excite me. My favourite thing to do with it is hitting 'randomise' on the blue components and seeing what happens. You get a brief look at that at around 1.54 in the video I posted. Absolute madness nearly always ensues, yet it rarely sounds 'ugly'. I'm still yet to try rendering all the decomposed components out for the granular fun that I alluded to earlier. I've barely set this new machine up at all so far tbh.

[WatchTheGuitar]

That looks like a lot of fun.

[Dirtgrain]

Does it ever go on sale?

[cron]

Does it ever go on sale?

Just had an email offering 20% off sitewide at Isotonik with the code BLACK2019

Valid until midnight GMT on the 30th of November.