Granular synthesis has a foundational assumption almost nobody has questioned in seventy years: a grain is a fragment of time. From Iannis Xenakis's acoustic particles in the 1950s, through Dennis Gabor's quantum theory of sound, through Curtis Roads's code in the 1970s and 80s, every granular instrument ever built has operated on the same premise — audio is material arranged along a time axis, and grains are small pieces of that timeline. You scrub forward, you scrub backward, you randomize position, you stretch duration, you build tonal clouds or rhythmic bursts. Ableton's Sampler, Max/MSP's Granulator II, Borderlands, countless others — all inside the same coordinate system. The time axis is infrastructure so deep it became definitional. Nobody thought to remove it.

Cristián Vogel's Incline removes it, and then asks what's left. Instead of assigning grain positions along a time axis, Incline uses machine listening algorithms to analyze your audio corpus, then arranges grains in two-dimensional space by perceptual similarity. Distance in the map doesn't mean "earlier or later in the recording" — it means "more or less acoustically alike." The contour lines drawn across the interface (Incline calls this a topographic map, and it genuinely looks like one) trace the latent topology of sonic content that has always existed inside every audio file, invisible and untouched, because the time axis always arrived first and organized everything before this topology got a chance. When you navigate the map — clicking across its terrain, letting the Contour Sequencer trace a contour line autonomously — you're traversing relationships that granular synthesis never made accessible.

In SAE terms: the time-axis coordinate system is the already-construct (已构). It has been chiseled so deep into the definition of granular synthesis that seventy years of granular innovation — randomization, pitch-shifting, time-stretching, density control — amount to parameter adjustments within the construct, never a challenge to the coordinate system itself. The time axis is not a feature of granular synthesis; it is the definition of granular synthesis. Incline's chisel-construct move is to remove the time axis from the definition entirely. What remains after that removal is the remainder (余项): the topological content relationships between grains — how they cluster, how they differ, what paths connect them — which have always been latent in every audio file but were never once made playable, because the time axis was always there first.

Vogel named this space with a word he had to invent: toposonic. Today that word has no external definition; only he uses it; only this instrument instantiates the concept. CDM's Peter Kirn, reviewing Incline, wrote that he almost hesitates to call it "granular synthesis" — because what it does and what that name implies have too large a distance between them. That hesitation is a structural signal: naming is beginning — Synth Anatomy wrote about it, KVR Audio has a discussion thread, CDM ran a hands-on review — but naming is not complete. The granular synthesis literature is vast. "Toposonic corpus explorer" is still an empty set, and Incline is its only member.

The window is not indefinitely open. Two forces are converging on this remainder. ML audio tooling is consolidating under the "AI audio = prompt → output" narrative, and sonic content topology will increasingly be absorbed into opaque generative infrastructure — a technical detail sealed inside the black box rather than a navigable space. Simultaneously, academic corpus synthesis research — FluCoMa, audio descriptor navigation tools — is beginning to name adjacent territory; once that naming hardens, the toposonic topology becomes a labeled subcategory rather than an unnamed remainder. Incline exists before either closure. The Windows version hasn't shipped. The logic is still growing. Seeing it now and seeing it six months from now are judgments of two different things.

Han Qin (秦汉) · @hqinjarsy · SAE Judgment & Aesthetics