You are here

Zynaptiq Adaptiverb

Resynthesizing Reverb Plug-in
By Paul White

Zynaptiq Adaptiverb

We’ve come to expect the unexpected from Zynaptiq, and their new reverb delivers just that!

In real life, reverberation is created when sound reflects and re-reflects from hard surfaces in the environment. The traditional approach to building a synthetic reverb approximates the way these reflections behave using recirculating filters; also popular are convolution reverbs, which use impulse responses recorded in an acoustic space to recreate that space in software. However, anybody who knows Zynaptiq will know that they never pitch up with anything even vaguely normal when it comes to plug-ins! No surprise, then, that their Adaptiverb incorporates an entirely new method of reverb synthesis, where the main reverb engine is described as ‘reflectionless’.

Unnatural Adaptations

On the face of it, Adaptiverb does much of what you’d expect a reverb to do, in that it adds decaying ‘tails’ to a sound to give it a sense of space and depth. However, its main reverb section does this in a profoundly different way from everyone else’s: the reverb tail is generated using a multitude of oscillators that track the pitches of the harmonics making up the original sound. Adaptiverb does include a separate and more conventional all-pass-filter reverb stage if you need it, but the main purpose of that stage is to add diffusion to what the main reverb engine produces. Unlike most reverb plug-ins, though, it has no independent section devoted to early reflections.

What’s the advantage in doing things this way? After all, if somebody has already invented the wheel, do you try to go one better by putting walking boots on the end of a set of spokes without good reason? The short answer is that Adaptiverb is not about emulating real rooms, but rather offering a flexible and musically creative effect. At present this comes at the cost of considerable CPU power and latency, but even that is being addressed: a revised version is imminent that is more CPU-efficient and includes a new low-CPU preview mode that sacrifices little in the way of sound quality. Adaptiverb supports Audio Units on Mac OS, plus VST and RTAS/AAX on both Mac OS and Windows, and is authorised online using an iLok account, either to a specific computer or a second-generation iLok key.

In that it doesn’t emulate specific spaces, Adaptiverb could in some ways be considered as a modern counterpart to the plate reverb — but Adaptiverb also has additional controls and features that turn it into it a hugely powerful sound-design tool. These take it way outside conventional reverb territory, and had me hooked within minutes. As the harmonic tails generated by Adaptiverb’s main reverb engine use oscillators rather than conventional feedback loops, the decay is free from any graininess, and when you put your surreal hat on to explore the sound-design possibilities, it can be coaxed into generating a wonderful array of spectral processing effects. Zynaptiq have also added something they call Harmonic Contour Filtering (HCF), which can do all manner of tricks including preventing the tail end of the reverb decay clashing if a radically pitched note or chord follows.

A Freeze button acts on the source sound to create drones, plus there’s an HCF Hold function to lock in the current filter state so pads and drones can be fasioned from virtually any input. You can also shift the pitch of the harmonics in the reverb tail or hold the harmonic content of a reverb created by one input signal and then apply that harmonic structure to the reverb for processing a totally different instrument. A Simplify function also allows the harmonic complexity of the oscillator-generated reverb tail to be reduced to create textural effects — something that can sound very other-worldly — and you can even invert the HCF spectrum to have the effect emphasise only to those parts of the spectrum that the input signal doesn’t inhabit. Oh yes, and you can apply pitch quantisation to the reverb tail selecting only the scale notes you wish it to occupy. If you think this sounds like a recipe for total creative weirdness, then you’re right, but Adaptiverb can also generate absolutely sublime ‘normal’ reverb.

Unfamiliar Territory

As Adaptiverb uses an unconventional kind of reverb-tail synthesis, it stands to reason that it might also come with some new and unfamiliar controls. Of course there are presets to give you a crash course in what is possible, but to get the best out of this creative vehicle, you have to learn how to drive it.

Step one in understanding Adaptiverb is to look at the signal path from input to output. The input stage includes an adjustable low-cut filter, variable pre-delay and a high-frequency processor called Air, which adds a kind of steamy brightness to the top end. Next is the Freeze section, which, as its name implies, allows for the freezing of the source sound — but this isn’t like the infinite reverb algorithms you may be familiar with. Instead, it freezes a slice of the dry input feeding into the reverb engine, in a way that sounds not dissimilar from the EHX Freeze pedal. What comes next is the oscillator-based sustain generator that Zynaptiq refer to as the Bionic Sustain Resynthesizer. Here, short-time fluctuations such noise and transients are filtered out and then the steady part of the signal is used to generate the harmonic tails that combine to create sustain. In essence, hundreds of separate oscillators are all made to follow the harmonic structure and phase of the input signal.

We then reach the secondary Reverb section. As Adaptiverb doesn’t set out to generate early reflections and has no conventional EQ, this has few of the parameters associated with a typical algorithmic reverb (though there are controls for damping and reverb size). Three algorithm choices are available: All-pass, Ray Tracing or HD Ray Tracing. In the context of reverb generation, ray tracing calculates the paths that sound would take from source to listening position in a modelled space. Adaptiverb’s ray-tracing model can, we’re told, render the cumulative result of 16,000 different audio paths from two ‘virtual speakers’ to the listener, but without having to compute all the individual reflections. This produces a very smooth reverb tail that can go toe-to-toe with any high-end reverb device, especially if you enable the HD option. Unlike a typical algorithmic reverb, however, there is no early reflections section.

The reverb tail generated by the ray-tracing Reverb section has a flat frequency response, as the necessary ‘colour’ comes from the post-filtering HCF stage and the pre-filtering Sustain section. For the same reason, the all-pass section also generates relatively little coloration. The signal going into the Reverb section can be taken either from before or after the Bionic Sustain synth section, or a mix of both can be used according to the setting of the Reverb Source control. The most musical ‘conventional’ reverbs, to my ears anyway, come from using a combination of the Bionic Sustain engine with some contribution from of the second Reverb in ray-tracing mode, but the all-pass filters provide a useful alternative texture.

Finally, the Harmonic Contour Filtering or HCF section, which processes the reverberated signal, also hosts a Hold function to lock the current filtering state. A small virtual keyboard allows the user to set notes to force the reverb tail to conform to a specific set of pitches, rather like setting the scale on a pitch-correction plug-in. At the output we’re back in more familiar territory; in conjunction with the Reverb Source and Reverb Mix controls, the Wet/Dry Mix control provides the necessary flexibility for combining the contributions of the two different reverb sections in both series and parallel modes. Note that the apparent reverb level can be very different on different presets, and a separate Wet Gain control is available to help balance this out.

Bionic Power

The plug-in window is split horizontally to delineate the three parameter subsections. The controls for the Bionic Sustain synth are on the left, while those for the Harmonic Contour Filtering are on the right, with an X/Y control dominating the centre. An optional Fine Tune view makes visible a number of more subtle controls, and you can enable tooltips that provide you with balloon help as you mouse over the controls. When tooltips are active, a faint overlay marks out the functional blocks of the GUI.

One axis of the X/Y Parameter Field adjusts the Bionic Sustain synth reverb decay time, while the other sets the reverb’s dry/wet mix going into the HFC section — which should not be confused with the plug-in’s overall output stage wet/dry mix. Where the host DAW allows MIDI to be routed directly to a processing plug-in (Logic users look away now!), all the main parameters of Adaptiverb can be controlled this way, and a MIDI Learn mode simplifies setting up. Conventional parameter automation is also supported.

Naturally you can store your own presets, but the manual points out that some aspects of a preset may be signal-dependent; for example, both the Freeze and HCF Hold buffer contents are stored within the presets, so may not behave as expected when applied to other sound sources. Indeed, some of the factory presets create drones with no need for an input, as they already include a ‘frozen’ sound.

Additional controls are accessible in Fine-Tune mode.Additional controls are accessible in Fine-Tune mode.

In Fine-Tune mode, additional controls for the Bionic Sustain synth include Interval and Richness, which work together, allowing oscillators to excite other oscillators that are close to but not exactly the same as the specified musical interval pitch to create a denser, more textural sound. Pitch Randomisation adds random pitch modulation to the oscillators used to create the reverb tail, while Diffusion affects the Bionic Sustain synth’s output in a subtle way by manipulating interference between its inputs and outputs. Sustain sets the decay time of the ‘oscillator’ tail generated by the Bionic Sustain synth while Simplify — in addition to reducing the number of oscillators — actually controls multiple parameters behind the scenes. Note that as Freeze captures the input to the reverb engine, the reverb parameters for either reverb section can be changed or automated while the input is kept frozen. The Freeze control can also lock the HCF settings to produce a static filter transfer function if the HCF is set to Track-Linked.

In this mode, the HCF tracks the harmonic content of the output of the reverb, compares it with that of the input signal, and suppresses anything that doesn’t match up. Positive values help to avoid clashes between the dry signal and reverb at chord or key changes; tracking can also be set to negative values to suppress matching frequencies, or it can be left in neutral. As touched upon earlier, Keyboard mode allows the effect to be filtered by key and scale; either resonant filtering or harmonic quantisation can be used in the filtering process. The former sets up a bank of steep band-pass filters at the frequencies of the harmonics relating to the pitches set with the keyboard, while the latter will quantise all harmonics to the note frequencies set on the keyboard. Keyboard settings can be stored as up to five snapshots, which can be called up on screen or automated within your DAW.

To Infinity... And Beyond?

Zynaptiq have created an absolutely huge range of factory presets, which are split into logical categories, from music production to sound design. It’s worth having a wander through these just to see what makes them work, and in many cases you can change a few parameters to customise them to your present situation. While the music-production reverbs sound wondrously lush and smooth, there are also ethereal spectral washes, spacey sounds, drones, virtual pads, chordal reverbs and pitch-shifted reverbs. The ability of this plug-in to take a normal sound and then twist it into something entirely new is beyond impressive — but then so is its capacity to take on the more usual reverb roles in an extremely mature manner, where the lack of ‘tearing cloth’ early reflections actually becomes very seductive. It does take an hour or so to find your way around the controls, but the journey is worth the shoe leather, and there aren’t too many adjustments to deal with in total — it’s more about learning how the controls interact.

On the one hand Adaptiverb is the kind of tool sound designers dream about, but its capacity for weirdness shouldn’t detract from the fact that it is also a fantastic studio reverb for vocal and instrument applications, with its unique ability to avoid pitch clashes where appropriate. I can honestly say that I’ve never come across a reverb plug-in that can offer so many different things to so many potential users, and the sheer quality of the results allows me to forgive it for its CPU load and latency.

Alternatives

You might get close if you buy some high end reverb plug-ins and some spectral effects tools, but to get all of this in one plug-in, plus unique features such as the ability to avoid the reverb tail clashing with the following note or chord, is unique.

Bionic Sustain Resynthesis

I’ll use mainly Zynaptiq’s own words to describe how the Bionic Sustain Resynthesizer works, as I’d hate to be responsible for any technical inaccuracies! They tell us that the design of the oscillator network used in this section is derived from an approach used in bionics to teach self-driving cars how to avoid obstacles.

In Adaptiverb, the process learns how to recreate the most significant aspects of the input sound. The approach is structurally similar to how artificial neural networks work, except that it uses a network of non-linear oscillators, trained using a Hebbian learning rule, as ‘knowledge nodes’, rather than as a network of simulated neurons. In a way you can think of the process as ‘re-imagining’ the input sound, with a longer sustain/release time, intelligently stripping out perceptually unimportant detail while retaining important signal features. So now you know!

Latency

The minimum latency achievable with Adaptiverb is 4096 samples, thanks to the Bionic Sustain Synth engine, which is always active. You’ll also need to set a large buffer size to ensure stable operation. Furthermore, using a ray-tracing rather than all-pass algorithm in the Reverb section increases the latency to 7168 samples, so worst-case scenario is a latency of 11,264 samples or around a quarter of a second at 44.1kHz.

While this is not an issue when mixing, it could be a real problem when playing live, so Zynaptiq have added a Live mode that reports zero latency to the host while disabling its internal dry path delay so that the dry path passes through latency-free. The reverb tail will still start a little late, of course, but where long reverb tails are being used, you may not notice.

Pros

  • Exemplary music-production reverbs.
  • Wide range of abstract effects.
  • Sensible number of controls.

Cons

  • Heavy CPU overhead.
  • High latency.

Summary

Adaptiverb is a single tool that can tackle both music-related reverb tasks and sound design. Its new approach to reverb generation makes it sound very smooth and musical.

information

€269 including VAT.

www.zynaptiq.com

Published November 2016