SUNDER amplitude splitter by Unfiltered Audio is released

[tiker01]

https://shop.propellerheads.se/product/ ... -splitter/

Sunder is the worlds first amplitude splitter available in any format and we’re extremely excited to debut this algorithm as a Rack Extension.

Inspired by the "multi-latch gating" effects pioneered by Toni Visconti, Sunder is the result of years of research, testing, and experimentation.

Product details
With Sunder you can split an audio source into low, mid, and high volume amplitude bands. Apply separate amounts of gain to each band in order to achieve unprecedented levels of control over dynamics. Additionally each band can be soloed/muted, allowing you, for instance, to only hear the medium volume portions of a drum track.

A real-time custom display shows you exactly what is happening under the hood. Each band is color coded to show you what is being sent where.

The real fun begins when you start applying different effects to the three amplitude bands. This enables you to create all kinds of nonlinear processing (a.k.a. amplitude-dependent sounds). You can instantly add an organic feel to synthetic instrument or spice up acoustic tracks with custom nonlinear signal paths that would be difficult or impossible to achieve in any other way. There are many recipes to choose from:

Apply different flavors of reverb to each band to create the impression that loud sounds are coming from farther away and vice versa.

Set up a nonlinear distortion that gets more aggressive on quieter material.

Only add EQ on the loudest hit of a drum loop.

Create hybrid effects, where the low band gets reverb, the mid band is filtered, and the high band goes through a delay.

In addition to being able to control and modulate amplitude cutoff points in real-time, you can also modify each aspect of the envelopes being used to transition between the bands. This gives you complete control over transients- all of which helps sculpt the overall sound. Sunder’s signature amplitude splitting algorithm is “flat-sum,” meaning that if you combine the output of Sunder’s three bands at any moment in time, you’ll be left with an unaltered version of your input signal (unless you apply gain to any of the individual bands of course). One of the more useful features is the “input gain compensation” switch which allows you to use the input gain knob to quickly push the signal higher or lower, changing the splitting result without effecting the output volume.

We’ve also ported some of the most popular features of G8 and Yoko into Sunder: for instance a fully-featured sidechain mode, lookahead control, and individual send/return jacks for low, mid, and high bands.

As with any new type of effect, we realize that you might need help learning how to use it which is why we’ve included tons of combinator recipes that will help to get you started.

Special thanks to Aerotronic who created all of the combinators marked with “-AT” (https://soundcloud.com/aerotronic)

Special thanks to Noel G. and EnochLight who have been painstakingly beta testing since 2015!

[mataya]

Wow, this really sound great! Thanks. Let's check that out.

M

[Aquila]

Hey, I'm living in a town called SUNDERland. Is this an omen?

[Marc64]

cool device

Skickat från min SM-G935F via Tapatalk

[Loque]

Hum... I am not sure about that... Gotta try it. I will test, how good this works when i use it in mid/side processing and with compressors and so on... But looks promising....

[normen]

Couldn't 100% make it out from the examples. This does work in the time domain doesn't it? Or does it work in the frequency domain like Logics Spectral Gate?

[RequiemMachine]

Wow. I didn't even know I needed this....lol

[Galaxy]

Couldn't 100% make it out from the examples. This does work in the time domain doesn't it? Or does it work in the frequency domain like Logics Spectral Gate?

From what I can make of it, it looks like this device works in the amplitude domain. Set an amplitude threshold at -12, than another at -22, everything above -12 is isolated and affected by the high level send/return, everything between -12 and -22 is isolated and affected by the mid level send/return, and everything below -22 is isolated and affected by the low level send/return.

It's a muliband amplitude splitter. The term multiband is deceiving tho, cause it doesn't rely on freq bands, rather amplitude thresholds.

So maybe best to refer to it as a Multithreshold Amplitude Splitter. That's at least what I made out of it.

[normen]

Couldn't 100% make it out from the examples. This does work in the time domain doesn't it? Or does it work in the frequency domain like Logics Spectral Gate?

From what I can make of it, it looks like this device works in the amplitude domain. Set an amplitude threshold at -12, than another at -22, everything above -12 is isolated and affected by the high level send/return, everything between -12 and -22 is isolated and affected by the mid level send/return, and everything below -22 is isolated and affected by the low level send/return.

It's a muliband amplitude splitter. The term multiband is deceiving tho, cause it doesn't rely on freq bands, rather amplitude thresholds.

That's at least what I made out of it.

The question was rather if theres a FFT involved, which is how most modern denoisers work. You convert to the frequency domain (i.e. do a FFT), take out the parts below threshold and then reconvert to the time domain (i.e. inverted FFT). This way you can remove for example high frequency noise while theres still high-amplitude low frequency content. Old denoisers only worked in the time domain so the absolute amplitude defined whats being removed (like a normal gate basically).

I think this one works in the time domain but I'm not 100% sure, thats why I'm asking.

[Galaxy]

Oh, sorry bout that. I misunderstood you.

It's a good question tho, and hoping that it does work in the time domain based on your explanation. That would make this device the first denoiser for Reason.

Trailing it would prove it does this or not. Correct?

[O1B]

Gorgeous. a la Roger Nichols D4 with Modern/Reason tweaks... Gain, FX..


D4 fan so I definitely will try this out. And, congrats to Unfiltered Audio for a beautiful device.

3 Stars? on day one?
The 'Hater'ation is getting Predictable. Diagnosis: Developer Envy.

[normen]

Oh, sorry bout that. I misunderstood you.

It's a good question tho, and hoping that it does work in the time domain based on your explanation. That would make this device the first denoiser for Reason.

Trailing it would prove it does this or not. Correct?

It would be if it worked in the frequency domain. Anyway I bought it (as I find it interesting either way) and it works in the time domain. I tested by having two synths, one playing a low-level noise and one playing an intermittent high-level note. Each time the note plays the low band is completely dead. So it basically works like a normal gate so to speak.

[tiker01]

Oh, sorry bout that. I misunderstood you.

It's a good question tho, and hoping that it does work in the time domain based on your explanation. That would make this device the first denoiser for Reason.

Trailing it would prove it does this or not. Correct?

It would be if it worked in the frequency domain. Anyway I bought it (as I find it interesting either way) and it works in the time domain. I tested by having two synths, one playing a low-level noise and one playing an intermittent high-level note. Each time the note plays the low band is completely dead. So it basically works like a normal gate so to speak.

So this device could be used to reduce some nose from old recordings after sampling?

[normen]

So this device could be used to reduce some nose from old recordings after sampling?

No. I mean - not better than any old gate. It DOES NOT work in the frequency domain. If you have high-amplitude low frequency content going in then the "low" output WILL NOT output the high frequency low-amplitude sound separately.

Anyway its not a bad thing. Processors that do this usually create a huge amount of "glass noise" (i.e. the noise typical for low-bitrate mp3 for example).

[O1B]

I think they mean SunderBAND

Hey, I'm living in a town called SUNDERland. Is this an omen?

[Loque]

Ok, here are my first impressions :
* nice, i like i
* the envelope needs a sync button
* i want to adjust the back blending of the return signal
* to see the difference between send and return volume woould be nice
* more detailed lookahead
* a bigger display, maybe with metering numbers (not a must)
* a latency compensation (ok, lol, impossible)
* higher gain values or remove those buttons completely

Really, the control over splitting and back blending would be awsome!

[Peter]

Wow pretty amazing concept! Looking forward to the trial this week!

[gullum]

Insta buy will try it when I get back from work

[JiggeryPokery]

3 Stars? on day one?
The 'Hater'ation is getting Predictable. Diagnosis: Developer Envy.

It almost certainly wasn't yet rated when you posted that: all new devices default to three stars. (Don't ask me why, they just are I guess it looks better than no stars)

[selig]

Couldn't 100% make it out from the examples. This does work in the time domain doesn't it? Or does it work in the frequency domain like Logics Spectral Gate?

If this works like the process I've been using since Reaktor first came out, it works in the DYNAMIC domain.

At it's simplest, think of a peak detector controlling a panner such that soft levels pan to one channel and the louder it gets the further it pans to the other. But instead of panning to left and right you simply sum the two panned signals. When you combine these two signals you get the input/unity (if it's done correctly), so it's only after modifying one of more of these signals that you get any 'effect'. What I described is two bands but I regularly use 3 or even 4 bands (my latest design is four bands).

Why do this? One thing I find very useful is to do things like simply raise the lower band, which smoothly brings up the level of all the low level signals without raising the peaks (a crude precursor of sorts to my Leveler). Of course you bring up the noise too, so that's one reason to use more than one band and have control over how the bands are split. Another thing I really like is simply brightening up the lower band, which adds brightness to ONLY the softer parts of the audio signal.

I've not tried this device so I'm more talking about my work rather than theirs. But I can only assume I'm not the only person to think of this idea, so I assume they are doing something very similar to what I've always done. And it CAN be a useful effect, but also not very obvious and easy to explain - all the engineers I've explained it to have a blank stare for the first few minutes, and this is one reason why it's not something I've yet released on my own (but it's definitely on "the list")!


Sent from my iPad using Tapatalk

[selig]

So this device could be used to reduce some nose from old recordings after sampling?

No. I mean - not better than any old gate. It DOES NOT work in the frequency domain. If you have high-amplitude low frequency content going in then the "low" output WILL NOT output the high frequency low-amplitude sound separately.

Anyway its not a bad thing. Processors that do this usually create a huge amount of "glass noise" (i.e. the noise typical for low-bitrate mp3 for example).

As I previously mentioned, I've been doing this with Reaktor for MANY years, and not experienced this effect. But I tend to change very little in any single band, and there are of course ways to minimize any issues once you spend some time with the concept.

What effects have you used that are similar to this?


Sent from my iPad using Tapatalk

[normen]

As I previously mentioned, I've been doing this with Reaktor for MANY years, and not experienced this effect. But I tend to change very little in any single band, and there are of course ways to minimize any issues once you spend some time with the concept.

What effects have you used that are similar to this?

Like I said, Logics "Spectral Gate" works in the frequency domain, basically all modern denoisers do as well. Theres been a few compressor plugins that also did it (dunno if the Roger Nichols thing did as well) but I never really liked the results.

You said that you did this in the time domain (where this plugin works as well) though? Or did I get that wrong? If you do it in the frequency domain you get the glass noise. Not sure if things get mixed up in our exchange here now. If you stay in the time domain theres no such issues.

To clear up any misunderstanding this is the difference between time domain and frequency domain:


Where "time domain" refers to manipulating the actual wave and "frequency domain" almost always implies using a FFT and then inverse FFT (or a filter bank if you want to go oldschool) and manipulating the single frequencies. "Dynamic domain" or "amplitude domain" doesn't really apply to this specific separation as you could go for amplitude in both frequency and time domain.

[synaptyx]

I love it already. Well done, Unfiltered Audio guys.

[Loque]

Anyone knows, if it has any advantage to feed the return back to Sunder instead of using a Mixer?

[selig]

As I previously mentioned, I've been doing this with Reaktor for MANY years, and not experienced this effect. But I tend to change very little in any single band, and there are of course ways to minimize any issues once you spend some time with the concept.

What effects have you used that are similar to this?

Like I said, Logics "Spectral Gate" works in the frequency domain, basically all modern denoisers do as well. Theres been a few compressor plugins that also did it (dunno if the Roger Nichols thing did as well) but I never really liked the results.

You said that you did this in the time domain (where this plugin works as well) though? Or did I get that wrong? If you do it in the frequency domain you get the glass noise. Not sure if things get mixed up in our exchange here now. If you stay in the time domain theres no such issues.

To clear up any misunderstanding this is the difference between time domain and frequency domain:


Where "time domain" refers to manipulating the actual wave and "frequency domain" almost always implies using a FFT and then inverse FFT (or a filter bank if you want to go oldschool) and manipulating the single frequencies. "Dynamic domain" or "amplitude domain" doesn't really apply to this specific separation as you could go for amplitude in both frequency and time domain.

Yes, time domain. It's like a multi-band crossover rotated 90°, affecting all frequencies equally. And as you suspected, audio is only in one or two bands at a time, unlike a multi-band device. It's more like a 3-4 way panner than a splitter. There are many analogies I've come up with over the years to try to describe this concept, since it's not a common idea in the audio world (which I won't go into unless asked). But once you understand it you see it's really quite simple basic concept IMO. Whether or not the concept is useful to anyone is altogether another question!


Sent from my iPad using Tapatalk