Adaptiverb uses an entirely new approach to reverb processing, and it sounds incredible. In Zynaptiq’s own words, “Rather than stacking delay/allpass nodes or convolving the input with an IR, Adaptiverb employs machine learning, ray-tracing and source separation techniques to synthesize a reverb tail that automatically adapts to the audio context it is used in.” As a result, Adaptiverb can produce less smearing and prevent discordant clashing of wet and dry signals when a pitched input signal’s harmonies change. Using Adaptiverb, you can also create convolved ambiences, filtered delay effects, static synth pads and more.
The cross-platform plug-in is available in AAX, AU, VST and RTAS formats. I reviewed Version 1.1.0 of the AU plug-in in Digital Performer 9.12, running OS X 10.9.5 on an 8-core Mac Pro.
Adaptiverb’s signal path comprises six processing blocks—some rather arcane—sequentially chained (with additional parallel signal paths) in the same order I’ll discuss them (see Fig. 1).
The Input Processing section contains controls for a low-cut filter, pre-delay and high-frequency synthesis called Air (the latter adds sparkle to the input signal). The Freezer processing block freezes the reverb’s input buffer (think infinite reverb). The Bionic Sustain Resynthesizer uses hundreds of adaptive oscillators that reproduce only the pitched parts of the input signal—thereby filtering out noise and transients—and generate a harmonic tail that can be adjusted using the plug-in’s Sustain parameter. The Reverb section provides additional diffusion based on either ray-tracing simulations or classic allpass algorithms; according to Zynaptiq, the ray-tracing engine simulates the effects of audio “taking roughly 16,000 different paths from two virtual speakers to the listener” and produces linear amplitude and decay characteristics. The primary purpose of the Harmonic Contour Filtering processing block is to eliminate dissonance between the input signal and reverb tail by removing discordant signal components from the reverb. But you can also “hold” (freeze) the HCF effect and apply it to another source to create cross-filtered effects similar to what you would achieve using convolution. After the HCF comes the plug-in’s Output section, which contains a global dry/wet mix control and a gain control for just the wet signal.
Some of Adaptiverb’s processing blocks are highly signal-dependent, and its affected Freeze and HCF Hold buffers are stored with presets. As a result, the plug-in’s output levels can vary dramatically, especially with percussive vs. steady-state input sources; I often needed to automate the wet signal’s gain control to compensate.
Adaptiverb’s intensive analysis engine and complex processing together incur latency between 4,096 and 11,264 samples. For use in performance, a Live mode provides a latency-free dry signal by reporting zero latency to your host (with the tradeoff that it also defeats automatic delay compensation).
Adaptiverb’s well-designed GUI is arranged in five sections: Meta (contains preset-management facilities), Input, Output, Reverb and Main Parameter, which has two views, Main and Fine-Tune (the latter offering an extended control set). All main parameters can be controlled by MIDI, subject to your DAW’s capabilities; MIDI Learn assignments are stored per preset rather than globally.
Unlike with conventional reverbs, the GUI’s Reverb section includes no controls for early reflections or frequency contouring; those controls either run totally contrary to Adaptiverb’s design (as with ERs) or are controlled adaptively by the HCF. The Reverb Model control selects one of three modes: Allpass (a dual-mono, traditional reverb that simulates rooms and halls), R-TRC (a 3D-raytrace simulation in true stereo) and R-TRC HD (a high-def variation of R-TRC mode). The Reverb Source slider controls the balance of signals fed to the Reverb block from the input and Freezer blocks (minimum position) vs. the Bionic Sustain Resynthesizer (max). You simultaneously adjust the size and decay time of the reverb using the Reverb Size slider. Change the reverb’s high-frequency damping using the Reverb Damp slider.
The reverb’s wet/dry mix (not the plug-in’s global wet/dry mix) is controlled by mouse-dragging a bright, blue dot along the horizontal axis of an X-Y field in the Main Parameter section of the GUI. The decay time for Sustain (the harmonic tail produced by the Bionic Sustain Resynthesizer) is controlled by dragging the dot along the vertical axis of the X-Y field. This construct blends the outputs of different processing blocks depending on how the Reverb Source slider is set. For example, if Reverb Source is set to its minimum value when you drag the Reverb Mix fully to the left in the X-Y field, the signal will be routed from the Input and Freezer blocks directly to the HCF (with no reverb applied). To help you grasp Adaptiverb’s complexity, a toggle overlays a block diagram of its alternate signal paths on the GUI.
In its stripped-down Main view, Adaptiverb’s Main Parameter section contains the X-Y field and Freeze button (in the GUI’s center), plus one basic control each for the Bionic Sustain Resynthesizer (left) and HCF (right). Unlike with other reverbs, the Freeze function’s infinite-reverb effect preserves the functionality of most other controls—allowing you to use it as a de facto synth-pad generator!
In the Main Parameter’s Fine-Tune view, expanded controls for the Bionic Sustain Resynthesizer respectively adjust the number of oscillator frequencies used for resynthesis, randomly modulate the oscillators, amplify existing harmonics for a richer sound (selecting octave, fifth or detuned unison intervals), and add diffusion.
Controls for the HCF either suppress similarities or dissimilarities between the input and effect signals; boost residual noise (useful for producing breathy synth pads with the Freeze function); adjust the timbre of the HCF’s effect; enable the HCF to track the Freeze function (retaining its current filter state); and provide a virtual keyboard that lets you choose fundamental pitches that will, depending on which of four accompanying algorithms are selected, have their harmonics resonantly bandpass-filtered or—with greater pitch discrimination—quantized by the HCF. You can store and recall five keyboard-setup snapshots.
On a haunting ballad, I fed Adaptiverb’s Bionic Sustain Resynthesizer to the R-TRC (ray-traced) reverb and used the virtual keyboard to select which of the singer’s notes would amplify long, ethereal reverb tails (while partially suppressing reverb on other sung notes). The shifting reverb timbres and levels produced an effect that was absolutely stunning, and unlike any I’d ever heard. Equally spectacular, I could use the HCF’s Hold function to freeze the filter state for the vocal reverb, save its buffer in a custom preset, and then apply its ever-shifting, voice-like timbres to Adaptiverb reverb on a keyboard track, thereby convolving it.
By clicking the Freeze button when the singer held a vowel sound, I produced a voice-like synth pad—albeit with a fixed pitch—that I could then spatially and harmonically alter by tweaking the HCF, Reverb Damp, Size and Bionic Sustain Resynthesizer controls. I could also create phase-y doubling and echo effects by plunging the reverb’s dry/wet mix to 100%-dry, killing the Bionic Sustain Resynthesizer’s output, cranking the HCF’s filtering, and dialing in pre-delays of various lengths.
One instance of Adaptiverb typically used 25% to 50% of my CPU resources, depending on my control settings; using an included Preview mode lessened the CPU load but incurred what I felt was an unacceptable tradeoff in sound quality on certain patches. Adaptiverb’s learning curve is fairly steep (luckily, Help balloons are provided) and there are no Undo/Redo buttons to use when you get off track. But Adaptiverb sounds so amazing, your efforts will be immensely rewarded. A portal to truly unique and exquisitely euphonic reverbs and intriguing special effects, Adaptiverb is a fantastic tool for the adventurous mix engineer and sound designer.
PROS: Unique. Sounds incredible. Can produce unusual effects in addition to awesome reverbs. Well-designed GUI.
CONS: Heavy CPU demand in high-resolution mode. Relatively steep learning curve. No Undo/Redo.