feat(studio): fold a preset shut, and give each one its own title design (#3191)

* fix(studio): close the typecheck and fallow gaps wa-18b-reschedule opened

useAutomationLanes.ts's write() assumed gesture-scoped coalescing and a
preview-only commit that useDomEditAttributeCommits.ts never grew — backported
that option support from its own later commit so the two sides of the API
agree. The paste path and its tests were missing the box selection's v0/v1
bounds a sibling commit added to AutomationSelection. The FX panel's carve
controls still edited the six mechanism numbers (maxCutDb, bands,
intelligibilityBias) after carveProfile() collapsed authoring to one Strength
knob, so those fields no longer existed on HfCarveSettings; UI now edits
strength, and analyseCarveBands is called with carveProfile(strength).

Also closes fallow's complexity, dead-code and duplication findings on this
PR's diff: extracted automationLaneDragMath.ts (pure group/point-move math)
and useAutomationRangeDrag.ts (the marquee-select gesture) out of
useAutomationLaneGestures.ts, pulled a couple of render-loop ternaries and a
resolver into named functions, dropped an export nothing outside its file
used, and shared a step-simplifier between audioCarve's two envelope
builders.

The edge-stretch vs. box-select priority test in TimelineAutomationLane.test
was still pinning the pre-box-select rule (edge wins over a point sitting on
it) that a sibling commit deliberately reversed — a point inside the box is
now selected content, so grabbing it drags the group instead. Updated the
test to the shipped rule instead of the old one.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix(core): cap the via conic's weight so an edge-clamped via point can't NaN

A via point pulled out past the segment (viaX: 5, viaY: -3) clamps to
(0.999, 0.001) — exactly on the steady region's edge, where edge - viaX is 0.
viaConic divided by that zero to get an infinite weight, and shapeVia turned
Infinity into NaN a few steps later (Infinity - Infinity in the quadratic
coefficient). NaN reaching setValueCurveAtTime silences the automated
parameter for the rest of the render.

Capped the weight at 1e6 instead of leaving it unbounded — past that point
the arc already reads as touching the via point, so nothing visible is lost.
Also hardened shapeVia's existing denominator guard (`<= 0`) to `!(> 0)`,
since NaN fails the original comparison and fell through it.

Review by Miga (PR #3208).

* fix(studio-server): fingerprint the proactive waveform cache key too

The route already keys the waveform cache on the asset's size and mtime as
well as its path, so a rebuilt-in-place file gets fresh peaks instead of
stale ones. generateWaveformCache — the proactive path that runs on upload —
still called buildWaveformCacheKey with the path alone, so it wrote to a
different key than the route reads from (making the pre-generated cache
never found) and kept the exact collision bug this fingerprint exists to fix
on its own path.

Review by Miga (PR #3211).

* style(docs): run oxfmt on the /hyperframes-audio skill docs

Table column widths had drifted out of alignment with oxfmt's own rules,
failing format:check and blocking the Preflight gate every downstream
branch inherits. Whitespace only, no content change.

* fix(core): stop \b from missing underscore-separated names, guard clipsOverlap's negative duration

\b treats `_` as a word character, so \bbed\b never matched bed_01,
music_bed_loop, or theme_song, and \bvo\b/\bvox\b/\btts\b had the same gap —
an underscore-separated bed classified as "unknown" and could end up offered
as its own carve source. Replaced the short hints with a boundary that
actually excludes letters and digits on both sides.

clipsOverlap computed end = start + duration without guarding sign, so a
negative duration put end before start — an interval that does not describe
anything, and one specific case showed it silently dropping a real overlap
(a shorter, earlier broken end rejected a clip that genuinely contained the
point). Duration clamps to zero instead: a clip cannot un-play time, and a
zero-length clip at its start is the sane reading of "duration nobody wrote
down as positive."

Review by Miga (PR #3212).

* fix(studio): widen PropertyPanel's resetModules render timeout again

The 20s margin (already once widened for the same reason) is timing out in
CI's full-monorepo Test run — the resetModules()+fresh-import render this
test needs is uncached and competes with every other package's test suite
for the same worker pool, and the same test passes in well under 2s
standalone. Went to 45s rather than re-tuning to whatever number happens to
clear the current CI load, since that number moves every time CI gains a
package.

* fix(studio): stop the single-candidate auto-apply carve firing twice

Two auto-apply effects both fire when sourceOptions.length === 1: the
multi-candidate effect only guards length === 0, so a single candidate
passes it too, and the single-candidate effect passes its own guard right
after — both compute the same sources list and both call setCarve, so the
common case (one narrator, one bed) triggered two decodes, two FFT runs, and
two concurrent attribute writes for one decision.

The multi-candidate effect now defers to its sibling for exactly one
candidate, which already has its own detailed handling for that case.

Review by Miga (PR #3213).

* feat(core): carve against every voice over a bed, always (#3212)

* feat(core): carve against every voice over a bed, always dynamically

A bed usually runs under a whole sequence — a narrator, an interview answer, a
second presenter — and carving against one of them left the others fighting it.
`source` becomes `sources`, and `mixCarveSources` sums every voice onto the BED's
clock before anything is measured. That is what keeps one analysis sufficient: the
chain is fixed, so there is no per-voice filter to switch between, and bands drawn
from all the speech there is with envelopes that rise wherever any of it happens
answer the actual question — where and when is speech masking this bed.

Summed rather than averaged: two people talking at once mask more than either
alone. Audio before the bed starts is dropped rather than folded in at zero, since
it plays over nothing and shifting it would put a cut where there is no voice.

`dynamic` is gone. A fixed depth thins the bed through every pause, and once both
have been heard there is no reason to want it, so every carve follows the speech.

Two helpers the panel and the headless script now share instead of each carrying a
copy — two definitions of "what does this name suggest" drift, and then the two
disagree about which track is the voice:

- `classifyAudioName` reads a track's kind from its id and filename together.
  `unknown` is deliberately common: treating an unrecognised name as "not a voice"
  would hide the one track somebody needs to pick.
- `clipsOverlap` keeps out a voice that never plays while the bed does. An unwritten
  duration counts as unbounded, not zero — refusing a clip whose length the
  composition leaves to the media would drop the commonest case there is.

Files written before this still load: a single `source` reads as a one-voice list,
a stored `dynamic` is ignored, and an absent attribute means the defaults whole.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix(core): stop \b from missing underscore-separated names, guard clipsOverlap's negative duration

\b treats `_` as a word character, so \bbed\b never matched bed_01,
music_bed_loop, or theme_song, and \bvo\b/\bvox\b/\btts\b had the same gap —
an underscore-separated bed classified as "unknown" and could end up offered
as its own carve source. Replaced the short hints with a boundary that
actually excludes letters and digits on both sides.

clipsOverlap computed end = start + duration without guarding sign, so a
negative duration put end before start — an interval that does not describe
anything, and one specific case showed it silently dropping a real overlap
(a shorter, earlier broken end rejected a clip that genuinely contained the
point). Duration clamps to zero instead: a clip cannot un-play time, and a
zero-length clip at its start is the sane reading of "duration nobody wrote
down as positive."

Review by Miga (PR #3212).

---------

Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix(studio): port the carve UI off the removed source/dynamic fields

#3212 (accidentally squash-merged into this branch instead of main) changed
HfCarveSettings from a single `source` + `dynamic` toggle to a `sources`
list with dynamic mode removed outright — the multi-voice UI consumer that
goes with that shape lands in the very next PR, so this branch was left with
a type that no longer matched its own code.

Minimal port, not the multi-voice redesign that PR does properly: the
"Listen to" picker and analyse() treat sources[0] as the one voice this UI
still understands, and every dynamic-mode branch (the automated envelope
lanes, the toggle, the checkbox) is gone along with the field — a carve is
now always the static value the analysis computes, matching what the type
change made permanent. Test suite trimmed the same way: the automation-lane
and toggle tests covered behavior that no longer exists.

* chore: fix markdown formatting

---------

Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>

Author: vanceingalls

packages/core/src/audio/audioFxGraph.test.ts

@@ -698,3 +698,66 @@ describe("a preset's run is wrapped in a wet/dry blend", () => {
     expect(live).toEqual([]);
   });
 });
+
+describe("a preset's wrap stays with the nodes it belongs to", () => {
+  const peak = (over: Partial<HfAudioFxNode>): HfAudioFxNode =>
+    ({
+      type: "peaking",
+      enabled: true,
+      params: defaultAudioFxParams("peaking"),
+      ...over,
+    }) as HfAudioFxNode;
+
+  it("rebuilds when a reorder moves a node across a preset boundary", () => {
+    // The type sequence is identical either way, so without preset identity in
+    // the shape the chain updated in place and left the wet/dry wrap bracketing
+    // the hand-added effect instead of the preset's — preview blending out the
+    // author's own node while the render, which rebuilds, blended out the
+    // preset's.
+    const before: HfAudioFxChain = {
+      version: 1,
+      nodes: [peak({ id: "p1", fromPreset: "boom-tame" }), peak({ id: "own" })],
+    };
+    const after: HfAudioFxChain = {
+      version: 1,
+      nodes: [peak({ id: "own" }), peak({ id: "p1", fromPreset: "boom-tame" })],
+    };
+    expect(buildFxChain(asCtx(ctx()), before).update(after)).toBe(false);
+  });
+
+  it("gives every run of one preset its own blend", () => {
+    // A preset pulled apart by a reorder occupies two runs. Keyed by id and
+    // assigned, the second wrap overwrote the first, so a whole-preset lane
+    // reached one fragment and the switch silently left the rest applied.
+    const split: HfAudioFxChain = {
+      version: 1,
+      nodes: [
+        peak({ id: "t1", fromPreset: "telephone" }),
+        { type: "reverb", id: "own", enabled: true, params: defaultAudioFxParams("reverb") },
+        peak({ id: "t2", fromPreset: "telephone" }),
+      ],
+    };
+    const built = buildFxChain(asCtx(ctx()), split);
+    // Two wraps, so two wet/dry pairs — four params under the one id.
+    expect(built.presets.telephone).toHaveLength(4);
+  });
+
+  it("pushes an amount into each run of a split preset, not one of them twice", () => {
+    const at = (amount: number): HfAudioFxChain => ({
+      version: 1,
+      nodes: [
+        peak({ id: "t1", fromPreset: "telephone", presetAmount: amount }),
+        { type: "reverb", id: "own", enabled: true, params: defaultAudioFxParams("reverb") },
+        peak({ id: "t2", fromPreset: "telephone", presetAmount: amount }),
+      ],
+    });
+    const built = buildFxChain(asCtx(ctx()), at(1));
+    expect(built.update(at(0))).toBe(true);
+    // Every wet leg off and every dry leg fully open: switching the preset off
+    // has to silence all of it, not the last fragment only.
+    const wets = (built.presets.telephone ?? []).filter((_, i) => i % 2 === 0);
+    const drys = (built.presets.telephone ?? []).filter((_, i) => i % 2 === 1);
+    for (const t of wets) expect(t.param.value).toBe(0);
+    for (const t of drys) expect(t.param.value).toBe(1);
+  });
+});

packages/core/src/audio/audioFxGraph.ts

@@ -599,7 +599,16 @@ function shapeOf(chain: HfAudioFxChain): string {
       // same reason: pushed into the running graph it would be a no-op, and
       // preview would keep sweeping on a triangle while the render used a sine.
       const wave = node.type === "phaser" ? `~${p.type}` : "";
-      return `${node.type}${poles}${fixedFreq}${wave}`;
+      // Which preset run this node belongs to, for the same reason again: the
+      // wet/dry wrap is WIRED around a run at construction, so moving a node
+      // across a preset boundary changes the graph's shape even when the type
+      // sequence is identical. Without this, reordering a hand-added peaking
+      // filter past a preset's peaking filter kept the in-place update path and
+      // left the wrap bracketing the wrong effect — preview blending out the
+      // author's own node while the render, which rebuilds, blended out the
+      // preset's.
+      const run = node.fromPreset ? `%${node.fromPreset}` : "";
+      return `${node.type}${poles}${fixedFreq}${wave}${run}`;
     })
     .join("|");
 }
@@ -674,8 +683,14 @@ export function buildFxChain(
 
   const shape = shapeOf(chain);
 
+  // Accumulated, not assigned. A preset pulled apart by a reorder occupies more
+  // than one run, and each run gets its own wrap — keying by id and assigning
+  // dropped every wrap but the last, so a whole-preset lane drove one fragment
+  // and left the rest at full strength while the switch read "Off".
   const presetTargets: Record<string, FxParamTarget[]> = {};
-  for (const p of presets) presetTargets[p.id] = mixTargets(p.wet.gain, p.dry.gain);
+  for (const p of presets) {
+    (presetTargets[p.id] ??= []).push(...mixTargets(p.wet.gain, p.dry.gain));
+  }
   return {
     input,
     output,
@@ -699,10 +714,14 @@ export function buildFxChain(
       // The blend is a value like any other: switching a preset off writes
       // `presetAmount`, and pushing it into the running graph is what keeps that
       // from being a rebuild — and from restarting the audio underneath it.
+      // Walked in step with the build, not looked up by id: `find` returned the
+      // first wrap for every run sharing a preset id, so a split preset wrote
+      // one wrap twice and never touched the other.
+      let wrapIndex = 0;
       for (const run of presetRuns(enabledAudioFxNodes(next))) {
         if (!run.preset) continue;
-        const wrap = presets.find((p) => p.id === run.preset);
-        if (!wrap) continue;
+        const wrap = presets[wrapIndex++];
+        if (!wrap || wrap.id !== run.preset) continue;
         wrap.wet.gain.value = run.amount;
         wrap.dry.gain.value = 1 - run.amount;
       }

packages/core/src/audioFxCopy.test.ts

@@ -103,3 +103,74 @@ describe("audioBandAt", () => {
     expect(audioBandAt(Number.NaN)).toBeUndefined();
   });
 });
+
+/**
+ * Copy that assumes the track is a voice.
+ *
+ * The rack sits on whatever the author selected — a music bed, a sound effect,
+ * a room tone. Every effect, every named job and every profile is offered on
+ * all of them, so a control that reads "Thins the voice out" on a synth pad is
+ * describing something the author cannot hear and does not have. It is the same
+ * rule this file's header already states — describe a control by what changes in
+ * THE SOUND — applied to the words rather than to the mechanism.
+ *
+ * The voice presets are the deliberate exception: they are voice by definition,
+ * and the shelf hides them on a track that classifies as music or as an effect,
+ * so "My voice sounds amateur" is only ever read next to a voice.
+ *
+ * This is a lint on the words, not a judgement about mixing. It exists because
+ * the offending strings were written one at a time over months and read fine in
+ * isolation — nobody notices the assumption until they apply Cut Rumble to a
+ * bass line and the panel tells them it will thin their voice out.
+ */
+describe("no copy assumes the track is a voice", () => {
+  /** Words that only mean something if the material is speech. */
+  const SPEECH =
+    /\b(voice|vocal|voices|speech|spoken|word|words|sentence|sentences|syllable|syllables|narration|narrator|talking|chest)\b/i;
+
+  /** Presets whose whole purpose is a voice, so their copy may say so. */
+  const VOICE_PRESETS = new Set(
+    HF_AUDIO_FX_PRESETS.filter((p) => p.family === "voice").map((p) => p.id),
+  );
+
+  const offenders = (entries: [string, string][]): string[] =>
+    entries.filter(([, text]) => SPEECH.test(text)).map(([where, text]) => `${where}: "${text}"`);
+
+  it("not in an effect's name, blurb, reach-for line or any knob", () => {
+    const entries: [string, string][] = [];
+    for (const [id, copy] of Object.entries(EFFECT_COPY)) {
+      entries.push([`${id}.title`, copy.title], [`${id}.does`, copy.does]);
+      entries.push([`${id}.reachFor`, copy.reachFor]);
+      entries.push([`${id}.primaryEnds.low`, copy.primaryEnds.low]);
+      entries.push([`${id}.primaryEnds.high`, copy.primaryEnds.high]);
+      for (const [key, param] of Object.entries(copy.params)) {
+        entries.push([`${id}.${key}.label`, param.label]);
+        if (param.hint) entries.push([`${id}.${key}.hint`, param.hint]);
+        if (param.ends) {
+          entries.push([`${id}.${key}.ends.low`, param.ends.low]);
+          entries.push([`${id}.${key}.ends.high`, param.ends.high]);
+        }
+      }
+    }
+    expect(offenders(entries)).toEqual([]);
+  });
+
+  it("not in the band vocabulary, which every spectral module shares", () => {
+    // These names get taught once and then reused everywhere, so a voice-only
+    // word here spreads to every filter in the rack.
+    expect(offenders(BANDS.map((b) => [b.name, b.says]))).toEqual([]);
+  });
+
+  it("not in the complaint a non-voice preset answers", () => {
+    const entries = Object.entries(PRESET_PROBLEM).filter(([id]) => !VOICE_PRESETS.has(id));
+    expect(offenders(entries as [string, string][])).toEqual([]);
+  });
+
+  it("still lets the voice presets say what they are for", () => {
+    // The exception has to be real, or the rule above is untested — a catalogue
+    // where nothing said "voice" would pass every assertion here vacuously.
+    const voiced = Object.entries(PRESET_PROBLEM).filter(([id]) => VOICE_PRESETS.has(id));
+    expect(voiced.length).toBeGreaterThan(0);
+    expect(voiced.some(([, text]) => SPEECH.test(text))).toBe(true);
+  });
+});

packages/core/src/audioFxCopy.ts

@@ -66,13 +66,13 @@ export const EFFECT_COPY: Record<string, EffectCopy> = {
     does: "Cuts the very bottom — traffic, footsteps, air conditioning, hands on the mic.",
     reachFor: "There's a low hum or thump under everything.",
     primary: "frequency",
-    primaryEnds: { low: "Only the deepest", high: "Thins the voice out" },
+    primaryEnds: { low: "Only the deepest", high: "Thins it out" },
     band: [20, 300],
     params: {
       frequency: {
         label: "Cut below",
         hint: "Everything under this is removed.",
-        ends: { low: "Only the deepest", high: "Thins the voice out" },
+        ends: { low: "Only the deepest", high: "Thins it out" },
       },
       q: { label: "Sharpness", hint: "How abruptly the cut starts." },
       poles: { label: "Steepness", hint: "How fast it falls away below the point." },
@@ -134,7 +134,7 @@ export const EFFECT_COPY: Record<string, EffectCopy> = {
   compressor: {
     title: "Even Out Loudness",
     does: "Brings the quiet parts up and holds the loud parts down, so nothing jumps out at the listener.",
-    reachFor: "Some words are much louder than others.",
+    reachFor: "Some parts are much louder than others.",
     primary: "strength",
     primaryEnds: { low: "Barely touched", high: "Very even, quite squashed" },
     params: {
@@ -168,10 +168,10 @@ export const EFFECT_COPY: Record<string, EffectCopy> = {
   },
   gate: {
     title: "Silence the Gaps",
-    does: "Mutes the pauses between words. Room tone under speech stays — this closes the silences, it does not remove noise.",
-    reachFor: "You can hear the room breathing between sentences.",
+    does: "Mutes the pauses. Whatever sits underneath stays — this closes the gaps, it does not remove noise.",
+    reachFor: "You can hear the room in the gaps.",
     primary: "strength",
-    primaryEnds: { low: "Only true silence", high: "Cuts quiet words too" },
+    primaryEnds: { low: "Only true silence", high: "Cuts quiet parts too" },
     params: {
       threshold: { label: "Quieter than this is a gap" },
       range: {
@@ -182,7 +182,7 @@ export const EFFECT_COPY: Record<string, EffectCopy> = {
       attack: { label: "How fast it opens" },
       release: {
         label: "How fast it closes",
-        ends: { low: "Clips word endings", high: "Leaves tails intact" },
+        ends: { low: "Clips tails short", high: "Leaves tails intact" },
       },
       knee: { label: "How gradual" },
     },
@@ -286,11 +286,11 @@ export const EFFECT_COPY: Record<string, EffectCopy> = {
  */
 export const BANDS: { from: number; to: number; name: string; says: string }[] = [
   { from: 20, to: 80, name: "Rumble", says: "traffic, footsteps, handling" },
-  { from: 80, to: 250, name: "Weight", says: "chest, body, warmth" },
+  { from: 80, to: 250, name: "Weight", says: "body, warmth, low end" },
   { from: 250, to: 600, name: "Mud", says: "boxy, muffled, cardboard" },
-  { from: 600, to: 2000, name: "Middle", says: "the body of a voice" },
-  { from: 2000, to: 5000, name: "Presence", says: "consonants, intelligibility" },
-  { from: 5000, to: 10000, name: "Edge", says: "sibilance, harshness" },
+  { from: 600, to: 2000, name: "Middle", says: "the body of the sound" },
+  { from: 2000, to: 5000, name: "Presence", says: "definition, consonants" },
+  { from: 5000, to: 10000, name: "Edge", says: "harshness, sibilance" },
   { from: 10000, to: 20000, name: "Air", says: "sparkle, openness" },
 ];
 
@@ -318,15 +318,16 @@ export const PRESET_PROBLEM: Record<string, string> = {
   "voice-broadcast": "I want it to sound like radio",
   "voice-warm": "I want it intimate and close",
   "rumble-cut": "There's a hum or thump underneath",
-  "room-gate": "I can hear the room between sentences",
-  "boom-tame": "My voice sounds boomy",
+  "room-gate": "I can hear the room in the gaps",
+  "boom-tame": "It sounds boomy",
   "harsh-tame": "It's harsh and tiring to listen to",
   telephone: "Make it sound like a phone call",
   "radio-am": "Make it sound like an old radio",
   megaphone: "Make it sound shouted through a horn",
   "lofi-tape": "Make it sound like an old tape",
   "pa-system": "Make it sound like a station announcement",
   intercom: "Make it sound like a door intercom",
+  "doofus-worble": "Make it wobble like it is seasick",
   "room-tight": "It sounds dry and stuck to the speaker",
   "room-natural": "It should sound like a real place",
   hall: "It should sound far away and big",

packages/core/src/audioFxJobs.test.ts

@@ -61,3 +61,20 @@ describe("named jobs", () => {
     for (const id of ids) expect(getAudioFxJob(id)?.id).toBe(id);
   });
 });
+
+/**
+ * A job's `does` is the complaint that leads to it, shown in the add menu on
+ * whatever track is selected — so it is under the same rule as the effect copy
+ * in `audioFxCopy.test.ts`: it may not assume the material is speech. Reduce Mud
+ * is as right on a boxy guitar as on a boxy voice, and the menu should say so.
+ */
+describe("no job assumes the track is a voice", () => {
+  const SPEECH =
+    /\b(voice|vocal|speech|spoken|word|words|sentence|syllable|narration|talking|chest)\b/i;
+  it("names the symptom without naming the source", () => {
+    const bad = HF_AUDIO_FX_JOBS.filter((j) => SPEECH.test(`${j.label} ${j.does}`)).map(
+      (j) => `${j.id}: "${j.does}"`,
+    );
+    expect(bad).toEqual([]);
+  });
+});

packages/core/src/audioFxJobs.ts

@@ -51,7 +51,7 @@ export const HF_AUDIO_FX_JOBS: readonly HfAudioFxJob[] = [
   {
     id: "tame-boominess",
     label: "Tame Boominess",
-    does: "Too much chest — it booms.",
+    does: "Too much low-end body — it booms.",
     type: "peaking",
     params: { frequency: 200, gain: -4, q: 1.4 },
   },
@@ -72,7 +72,7 @@ export const HF_AUDIO_FX_JOBS: readonly HfAudioFxJob[] = [
   {
     id: "add-clarity",
     label: "Add Clarity",
-    does: "Words are hard to make out.",
+    does: "It is hard to make out — it sits back.",
     type: "peaking",
     params: { frequency: 3000, gain: 2.5, q: 1 },
   },