Metabolism: song as substrate, ecology grown from the track

A new top-level section building the metabolism concept: the track supplies the food field, event impulses and per-section laws; the seed picks the species (a rule set, not a parameter vector). Same song + seed => bit-identical frames; different songs => different worlds. - analysis: single FFT pass -> per-frame feature table; score (onsets folded into world geometry via voice ring + pitch-class sector), wells (food placed by the harmony), graph (band correlations -> interaction matrix), laws (per-section regimes) - sim: typed-array world, 3 castes (builder/orbit, forager/commute, spark/orbit-wide) with causality rings, persistence trails, consequence field; fully deterministic - lens: software framebuffer renderer (PNG in node, ImageData in browser) + event-reactive camera (cuts on activity change, alternating shot sizes, drift-and-push) - metrics: coupling x residual fitness (not per-frame descriptors) plus a grown-structure fingerprint; report + strip tools; 15 tests green - DESIGN.md captures the concept, invariants and build order 3 reference test signals (drone/runner/dusk) grow measurably distinct worlds while seeds of one song stay in the same biome.
2026-10-24 08:23:33 +02:00
parent 6a2af2e45f
commit 13b0939221
26 changed files with 4469 additions and 0 deletions
@@ -0,0 +1,59 @@
+// Onsets become places. The song does not move the agents directly; it puts
+// food down, and the agents grow on it.
+//
+// Three rules make the geography legible:
+//   - the voice picks the ring: kick near the centre, body in the middle,
+//     air at the rim, so a track's mix is visible as an annulus;
+//   - the dominant pitch picks the angle, so twelve pitch classes become
+//     twelve sectors and harmony becomes direction;
+//   - strength sets how much food and how wide, so a hit is a place and a
+//     fill is a region.
+
+const TAU = Math.PI * 2;
+const RINGS = [0.24, 0.54, 0.82];
+
+export function buildEvents(analysis, rng) {
+    const events = [];
+    const keys = analysis.summary.chroma ? 12 : 0;
+    for (const o of analysis.onsets.all) {
+        const f = o.frame;
+        const hz = analysis.dominantHz[f] || 440;
+        const semis = Math.round(12 * Math.log2(hz / 440));
+        const pc = ((semis % 12) + 12 + 9) % 12;
+        // A quiet, noisy band has an unstable dominant bin. Pull the angle
+        // toward the event's own hash when the band is flat.
+        const confidence = 0.55 + 0.45 * (1 - (analysis.flatness[f] ?? 0.3));
+        const angle = ((pc / (keys || 12)) * TAU) * confidence +
+            rng.range(-0.16, 0.16) +
+            (o.voice - 1) * 0.05;
+        const radius = RINGS[o.voice] * (0.82 + 0.36 * rng.next());
+        const strength = Math.max(0, Math.min(1, 0.18 + 0.82 * o.strength));
+
+        events.push({
+            frame: f,
+            t: o.t,
+            voice: o.voice,
+            strength,
+            x: 0.5 + Math.cos(angle) * radius * 0.5,
+            y: 0.5 + Math.sin(angle) * radius * 0.5,
+            // World units. A wide event is a region, not a point.
+            spread: 0.014 + 0.040 * strength * strength,
+            // Food units. Scaled so a dense arrangement supports roughly a
+            // thousand agents and a sparse one a few hundred; see the report.
+            amount: (10 + 34 * strength * strength) * 0.07,
+            // Strong events seed life as well as food.
+            spawn: strength > 0.62 ? (o.voice === 0 ? 3 : o.voice === 1 ? 2 : 1) : 0,
+            born: 0,
+        });
+    }
+    return events;
+}
+
+/** Frame -> events, so the world steps in O(events this frame). */
+export function indexEvents(events, frames) {
+    const index = Array.from({ length: frames }, () => []);
+    for (const e of events) {
+        if (e.frame >= 0 && e.frame < frames) index[e.frame].push(e);
+    }
+    return index;
+}