UVI Focus Logo
Deep Dive
The R&D behind UVI effects

UVI has a long history of research and development in effects plugins. For over 10 years, our development team has worked to offer the most accessible products at the cutting-edge of audio processing. Let's take a look at how our team has designed six of the most iconic effects of our collection: Plate, Sparkverb, Dual Delay X, Phasor, Opal, and Drum Replacer.


One of our main goals for Plate, was to have a fully parametric plate reverb to be used creatively, by letting people design their own custom plate reverb way beyond what's possible in reality. Plate employs realtime physical-modeling with up to 20,000 modes to achieve new levels of depth and detail, going well beyond the limitation of physical units.

Physical modeling in action

The most iconic plate reverb, the EMT140, is a heavy device made of a large sheet of steel measuring 2m x 1m, with fixed transducers attached to the metal, and foam to control the amount of damping.

In contrast, thanks to physical modeling, Plate allows us to change:

• Physical dimensions for example a thin and long plate will sound similar to a spring reverb with very dispersive wave propagation of pulses like in railway
• Material, to change propagation speed and inharmonicity
• Tension, to control modal frequencies tuning
• Transducer placement, to change the reverb timbre


All kinds of damping are based on academic research and physically grounded, but we also made sure that one can play, extrapolate, and mix them intuitively in an easy-to-use interface with just a few clicks.


We even included some out-of-this-world features, like mode modulation in order to bring the lush chorused tails of algorithmic reverbs into the world of physical modeling.

We wanted Plate to be an authentic and accurate emulation. For that purpose, we went to a famous studio in Paris to perform in depth measurements on the well known EMT140 (steel) and EMT240 (gold foil). We then spent time calibrating the physical model to closely match our measurements for all kinds of settings.

We even included some out-of-this-world features, like mode modulation in order to bring the lush chorused tails of algorithmic reverbs into the world of physical modeling.

We wanted Plate to be an authentic and accurate emulation. For that purpose, we went to a famous studio in Paris to perform in depth measurements on the well known EMT140 (steel) and EMT240 (gold foil). We then spent time calibrating the physical model to closely match our measurements for all kinds of settings.

At UVI, the journey of Sparkverb was born from the internal need to create a fast, dense, efficient and high-quality reverb that passed the synthesizer-test, i.e. excelling with synthetic sounds. Departing from conventional approaches, we wanted Sparkverb to redefine reverb by offering a parametric design in order to continuously morph across a spectrum of spaces, from very small to very big, all within a concise parameter set, instead of having different modes like hall, room, chamber, or plate.


The challenge behind the development of Sparkverb was also to demystify reverb preset design, which can sound like black magic, or a task reserved to physicists for many people. So we pioneered the Preset Voyager interface. Leveraging machine learning, this interface learns the perceptual distances between factory presets and maps preset space onto a 2D surface, reflecting their perceptual similarity. Thanks to triangulation, one can navigate the preset space by continuously interpolating between nearby presets.


In reverb design, shaping the decay curve is crucial. Sparkverb introduces an innovative graphical interface that empowers users with direct, intuitive and playful visual interaction with the decay curve, transcending traditional knob-based adjustments.

When we started the process of designing Dual Delay X, our goal was to explore the sonic possibilities offered by the design constraint of two delay lines with feedback.

The foremost objective was to ensure that Dual Delay X remains user-friendly, allowing musicians to effortlessly reproduce the classic uses of parallel and ping- pong delays with intuitive damping and feedback parameters. Beyond these foundational use cases, Dual Delay X introduces an innovative feature: soundfield rotation between each feedback tap.

Dual Delay X

The concept of soundfield rotation opened doors to explore new categories of delay effects:

• Slow or fast rotating echoes, that allow dynamic echoes without relying on LFO modulation
• Inharmonic Comb Filtering: that allows users to experiment with unique, inharmonic textures by using shorter delays
• Scattering when the two delay lines are slightly different


To spice things up, we incorporated additional ingredients to shape the signal path within Dual Delay X as diffusion, dispersion, degradation and tape saturation.

Dual Delay X
The integration of these elements required us to create innovative algorithms, particularly for diffusion within the feedback chain, as the traditional ways of adding diffusion tend to accumulate into metallic resonances.
Our dedicated approach ensures that Dual Delay X delivers exceptional performance, with an all-new diffusion algorithm tailored to minimize those when used within a feedback structure.
Dual Delay X

To grasp what Phasor does to your sound, understanding phase shifting is key. Phase in audio refers to a waveform's position. By delaying the peak of a waveform, a phase shifter effect begins. Shifting the waveform's phase involves delaying it and then mixing it with the original, creating phase artifacts. This desynchronization continually generates new phasing effects. Many phase shifters let you adjust delay length and feedback to enhance the swirling effect. Some also offer filtered phased signals for greater control over audio artifacts.


UVI Phasor was first developed as an insert/send effect for Falcon. Running a variety of synths and samples through Phasor allowed our users to create some really cool tracks, and we quickly received requests to develop a standalone version of the effect that could be used in any DAW and on any sound source.

Phasor gives you an incredible amount of control over the phase shifting effects and audio artifacts that are produced. The engine itself is the timeless phaser engine that users of phase shift effects have come to know and love. It’s in the control and extra features, however, that Phasor really sets itself apart and shines.
The visual design of Phasor also allows you to have a real guide to how the controls you are adjusting will affect the sound. All of these controls and visual elements combine to allow the user to shape Phasor to the perfect settings in record time.
Opal is our brand-new optical program-adaptive leveler, designed to emulate a beloved vintage optical compressor from the ‘60s. We spent more than a year on research and development to create an emulation based on a physical model of the original circuit design. Opal reproduces the inherent program-dependent response times and nonlinearities which contribute to this famous compressor’s sonic signature.
The circuit
The circuit schematic was thoroughly analyzed stage-by-stage, including: gain reduction, sidechain and make-up circuit. As it contains about 40 electronic components, a brute force simulation would be too CPU-intensive for real-time use. Fortunately, some simplifications were possible without affecting the overall sound. In particular, we found we could differentiate between actual features of the circuit, and what mostly reflects hardware design constraints of the time, which we could then work around in a virtual analog model. Some stages were decoupled, some components were lumped together, and some components were forwent altogether.
The measurements
Next we performed measurements on several optical compressors to extract the characteristics of their optical cells, namely the photoresistor’s internal dynamics, and the law for the optical coupling between the photoresistor and the light-emitting element. As these were real machines used in real studios, the measurements had to be as non-intrusive as possible, meaning no circuit dismantling to measure components in isolation. Knowing the rest of the circuit’s electronics already, we could do some reverse engineering and estimate our model’s parameters for the optical cell with a bit of machine learning, based on these measurements.
The integration

For the simulation itself, we applied some reduction and pre-resolution techniques, and were able to reduce the model to 2 x 4th order systems (out of 40 components to begin with). The solver was also specially designed to be computationally efficient. Finally, we added modern features including external sidechain, variable responsiveness, frequency response correction, and tube drive, to create our own innovative physical model.


For Drum Replacer, we wanted to create a drum replacement tool, simple on the surface and easy to use, keeping the well established metaphor of level-based trigger detection. But under the hood, we spent time on a number of advanced signal cleaning and pre-processing methods, such as source separation and machine-learning methods to simplify the detection task.

Transience-based gating

Transience-based signal gating, while rather simple to use, can provide a considerable amount of track cleaning – it only lets transients pass through the detection module in a level-independent way, compared to traditional gating where setting a threshold is always governed by a compromise between missed events and false triggers. With Level-independent transience gating even weak events close in level to background noise can be isolated and detected.

Drum source separation
Our rationale regarding drum source separation was as follows: rather than training an average model on big datasets, we decided to use the most available information with local ad hoc drum models tailored to each specific drum recording situation. In most multi-mic drum set recordings there is a dominant drum kit element per microphone, sometimes polluted by bleed from other drumset elements. Using blind source separation methods, it is then possible to learn ad hoc models of the dominant drum layers, and then isolate and resynthesize only the main drum element we are interested in, simply by listening and toggling the different components.
Drum Replacer
Drum Replacer
Since signal is additively resynthesized in a constructivist manner, background bleed elements (e.g. hi-hats on a kick drum track) or noise can be efficiently removed. Combining these tools, borrowed from academia, provides a way to address difficult use cases by making the most of the available information about the problem at hand.

UVI Focus