Madrona Labs Sumu

[Ou_Tis]

How is the MPE integration? Do most of the presets already have Pressure/Aftertouch and Slide/Y axis/MPE Timbre already set up?

[dayjob]

via the madrona labs forums.

[kraster]

I wonder if the intro pricing is reflected in the bundle? If the bundle pricing doesn't change, its not a big deal to wait. Only $14 more for me to complete the bundle. The cpu tho, ouch...

The bundle situation is weird.

I have all the other Madrona plugins. Kaivo, Aalto, Virta and Aaltoverb but it only counts Kaivo and Virta in the bundle so the price for completing the bundle is €196 which is more expensive than Sumu on its own!

Randy said that the bundle pricing "doesn't work like that" on the forums to which I say, huh?

What's the point of it then?

Lots of companies add "loyalty" discounts based on what you already own. For instance, Sugarbytes and Minimal Audio.

[pekbro]

I wonder if the intro pricing is reflected in the bundle? If the bundle pricing doesn't change, its not a big deal to wait. Only $14 more for me to complete the bundle. The cpu tho, ouch...

The bundle situation is weird.

I have all the other Madrona plugins. Kaivo, Aalto, Virta and Aaltoverb but it only counts Kaivo and Virta in the bundle so the price for completing the bundle is €196 which is more expensive than Sumu on its own!

Randy said that the bundle pricing "doesn't work like that" on the forums to which I say, huh?

What's the point of it then?

Lots of companies add "loyalty" discounts based on what you already own. For instance, Sugarbytes and Minimal Audio.

I dont have the reverb and it showed me $151 to upgrade.

[kraster]

I'm very intrigued by it but price is a little too spicy for me particularly in its current state.

The AU version doesn't work at all. The scale and unison menus are greyed out as well as the tilt and blend controls (that could be user error but if it is it's definitely not clear to me how to change it)

Also deleting cables is really janky.

I'm never a fan of buying things with the promise they'll be better someday.

And with all due respect to Madrona (I really love their synths) they're not exactly known for their expedience...

[aMUSEd]

Yes that's true - there still are no VST3 versions of Aalto and Kaivo

This comes as VST3 I hope?

[donato]

I'm very intrigued by it but price is a little too spicy for me particularly in its current state.

The AU version doesn't work at all.

Working fine here in Logic with AU.

[foosnark]

It's about 10% cpu per voice of polyphony for me (Ryzen 5 3600, Windows, Bitwig)... which honestly is fine by me given the kinds of busy sounds I've been making with it.

[masterhiggins]

Doesn't say that the dynamic discount isn't working. He just said it's unclear.

[Borbolactic]

I can't make head nor tail of this thing.

Everything sounds like trash can full of ferrets being rolled down a concrete stairs.

I should have a folder of presets in the final release along with several other beta testers...
[YT video link removed for this quote]

Thanks for your examples...

What might be nice is a before and after of the resynthesized Vutu sample...

Currently, it's hard to tell how, and ultimately perhaps, how well Sumu is affecting the sound after Vutu resynths it-- like what's the original and/or resynthesized sound, what does it sound like-- and, for example, offering something truly unique and ear-grabbing than, say, Izotope's Iris 2 or some other additive, spectral, granular or waveform-morphing synth.

Sumu seems to have great potential but I just don't hear it, yet, but maybe it's just my ears.

Also, WRT Sumu's UI, it sort of looks like Sumu's just scanning along the partials. IOW perhaps, the partials, wherever they are or might be, don't seem to be bouncing up and down and left and right beyond how they arrived. It all looks quite static/sampley, save for maybe the 'bubbles' ('Pulses') coming up or the '3D binaural' stuff and final output.

[mike_o]

Has anyone figured out how to import the .json files created by vutu into sumu, I can't get it to import no matter what I try.

[foosnark]

Also, WRT Sumu's UI, it sort of looks like Sumu's just scanning along the partials. IOW perhaps, the partials, wherever they are or might be, don't seem to be bouncing up and down and left and right beyond how they arrived. It all looks quite static/sampley, save for maybe the 'bubbles' ('Pulses') coming up or the '3D binaural' stuff and final output.

It's using the partial map as three different but related data sources (pitch, amp and noise) with 64 channels each. (I'm not sure how Vutu divides the noise between channels.)

Where typical spectral resynthesis would use 64 sine waves and a noise source, the osc section here is 64 2-op FM (or AM) pairs, with noise sources for both carrier and modulator. And you have control over not just the pitch, amplitude and noise of each (through 64-channel inputs) but the modulation index, ratio, and a linear offset to the modulator. You can also scale the pitch input so you can bend all of those partials inharmonically, or even flatten them to unison. Or use no pitch input on the carriers, but have the partial map's pitch data control modulator frequencies instead... and so on.

The time index on the partial map is another 64-channel input which you can control with the Pulses section or Envelopes section if you like -- it's very easy to get spectral delay-like effects and more unusual movement that way.

All of the inputs and outputs in the patcher have 64 channels, including the Envelope gate input (you can trigger envelopes with the Pulses section or amplitude data from the partial map... stuff like that is why it's nice to have additional "normal" envelopes and LFOs in the DAW to control things with).

[kraster]

Also, WRT Sumu's UI, it sort of looks like Sumu's just scanning along the partials. IOW perhaps, the partials, wherever they are or might be, don't seem to be bouncing up and down and left and right beyond how they arrived. It all looks quite static/sampley, save for maybe the 'bubbles' ('Pulses') coming up or the '3D binaural' stuff and final output.

By default with the time parameter connected to the +input it just scans through from left to right.

If you patch the Partial map's amp output to the oscillators amp input there are 64 controls signals being sent to the amp input on the oscillator section derived from the partial map. Same goes for the pitch and noise.

So with these basic connections along with an envelope for the Gates section you can reproduce what Vutu has produced.

It's the oscillator that makes the noise in the synth. The oscillator is actually 64 voices which can be FM or AM.

But the fun begins when you can decouple the partial map time component from its pitch or vice versa.

The pulses section generates 64 control signals that can be used to alter the other parts of the synth.

How the pulses are generated can be controlled by various functions in the probability section. The "map" of the pulse can be changed to different shapes, they can be snapped to the grid, you can offset higher partials generation rate, you can alter the centre of the pulses map, you can add noise to it etc. etc.

These signals are then output by the "shapes" section which can have different envelope shapes, rhythmic components or amplitudes.

You could say, for example plug this into the multiplier in the position of the partial map so now the position of each partial in the partial map is scattered in time. Or plug the pulses output into the mod index input to have randomy generated FM or AM components.

The spaces module allows you to spread out the partials in space with various different fields that you can alter.

After scratching my head with it for a while I get the point of it now.

Generally the issue with resynthesis is that you end up with a ton of parameters which is really unwieldy and a PITA to work with. It's not practical or particularly useful to be tweaking the pitch and amplitude and noise components of 64 oscillators

Sumu addresses that by consolidating all of that data into simple single cable streams each containing 64 components and then allows you to alter that data in various ways.

[Borbolactic]

Thanks, guys. I guess time will tell as more get a handle on Sumu and show more of what it's capable of.

----

"Generally the issue with resynthesis is that you end up with a ton of parameters which is really unwieldy and a PITA to work with. It's not practical or particularly useful to be tweaking the pitch and amplitude and noise components of 64 oscillators..." ~ kraster

I guess that's where mathematical functions come in to automate the process?

[gExpectations]

its definitely something and quite interesting. i havent fully made up my mind about it. i dont think they do a good job explaining the thing. top is basically midi control, 3d data stream derived from sample, envelope array, trigger array and bottom is 2op fm, vca, spatialization, filter, output stage. or in another way: it has a lot of ways to make data streams but tonality comes from 2op fm.

all the modules make sense to me, especially the spatialization stage because it works with the 64 partials idea. the resynth stage also makes sense because it gives you a lot of variant data streams while the data streams themselves have a good amount of regularity in them. i also appreciate that they didnt add any effects. i feel just adding a reverb would have made a lot of presets sound more impressive but this is better, everyone has enough reverbs by now. intuitively i would have liked a more extensive osc section but maybe it would just result in chaos or spike the cpu use, so we cant have it.

i guess my main problem is that i cant tell how many useful sounds i will get out of it and having noise every minute doesnt help when trying to learn the thing but ill keep the demo around to see if its something i would like to purchase.

edit: also the small circles are i guess modulation attenuation but i cant always tell what they are connected to and there is no tooltip or pretty picture in the manual.