What will the next innovation in synthesis be?

Additive, subtractive, concatenative, waveguides, FM, granular, phase distortion, karplus-strong, LA, scanned, vector, wavetable… We have all these different ways of making noise, but we haven’t come up with a new one since the late 90s. Why is that?

The simple answer, of course, is obvious: interaction matters more than sound. See, even a drop dead basic synth running on a raspberry pi can sound phenomenal. If everyone’s happy with how it sounds, why change that? The focus has been on making complex synthesis techniques more intuitive and playable. Innovations in controllers (like the Osmose or Seaboard) allow for unprecedented levels of expression, bringing out new nuances from existing synthesis engines.

But, I’m a sound designer. I use synths to make realistic sounds, and largely don’t care about the interface as I’m not a player, but a tweaker. My modular synth’s only input is a single force sensing resistor to trigger sounds, the synth does the rest. People like me have been playing with the same drivel from 40 years ago hoping for new results. So, what’s next for us?

---

I think Microsounds is the next big thing. I’ve spoken about this concept before, carefully controlling tiny packets of sound to create something altogether new, unlike granular. The concept of manipulating sound at extremely fine temporal and spectral resolutions, as championed by Curtis Roads, is highly promising, and already seeing some limited use in devices like Wave Packets by Auza.

But, you can do that on a Moog or Serge or even a Roland JX. If we were to conjure up a “next big thing” in synthesis, I think it would likely revolve around “Perceptual Synthesis.” Not sure what to call it yet, but it’s a blend of existing and emerging technologies that would allow you to control vast numbers of microsounds on a macro level, like we see commonly with additive synthesis.

Instead of merely chopping and re-pitching audio, the system would allow for control of features like roughness, brightness, metallic, woodiness, vocal, sustained or transient, even emotion. Such a synth could even work like modern LLM’s, analyzing incoming audio (or a pre-existing sound library) and extract high-level perceptual features which the synthesis engine would then allow you to “dial in”. You wouldn’t turn an attack knob, but a softness knob, and it could change everything from the exact envelope of a given packet or overall sound, to the timbral quality.

Such an engine could even be controlled via natural language processing where you effectively prompt it for a sound like the, “sound of a small object rattling,” “the sound of liquid dripping,” or “the subtle scrape of fabric,” and it would generate it for you based on what it knows things sound like and how to shape microsounds to create a more cohesive sound. It would synthesize the individual elements of a sound, for example wind, air, friction, resonant frequencies, and insect-like transients, and combine them in a perceptually coherent way.

This tech could also be used for the same purposes additive resynthesis is used, but with much more control and potentially more accurate recreations of sounds, but with the ability to control individual elements literally down to the grain to allow for more complex changes to the initial sound than additive resynthesis can offer.

A possible control method for this type of synth could be something us modular folk have been dabbling with for decades: bio-inspired organisation. What if sound could evolve based on rules of life, aka how animals, plants, and insects behave? These interactions would lead to emergent, complex soundscapes that evolve and change over time in seemingly unpredictable but often beautiful ways. Such is the way with generative modular patches, which are often based on a chaos source like a Lorenz attractor or, as recently popularised by Instruo’s Scion, plant life.

---

We could call it Morphogenetic Synthesis, actually, based on principles of biological growth and form. Using many of modular’s generative sequencing techniques, sounds tend to grow from a single seed, like a clock, LFO, etc. Implementing that into the synthesis engine itself, rather than just as a sequencer, would allow for developing resonant structures, harmonic relationships, and temporal patterns that mimic natural development. This would lead to sounds that feel incredibly organic and alive – and sound incredibly unique.

Of course, in modular we already have this level and style of control. Paired with a microsound-based synthesis engine alone it could be powerful, but I feel adding the natural language processor would not only allow for more organic changes than editing single parameters at a time, it would allow such a powerful synthesis method to be used by pretty much anyone.

What do you think? What will be the next major change in synthesis?