Clarify voice patch target controls

state/evidence.jsonl

@@ -127,4 +127,4 @@
 {"date":"2026-05-20","topic":"weksa-v01-training-artifact","lesson":"The Weksa blocker has moved from schema definition to artifact consumption. Weksa now emits a machine-readable `weksa.utterance_embedding_handoff.v0.1` tiny PanPhon seed batch for `a`, `pa`, `ta`, `ka`, `sa`, and `ma`; AquaSynth's next serious training pass should load that artifact rather than local toy vectors.","evidence":"Added Weksa emitter `E:/Projects/weksa/tools/emit_tiny_panphon_handoffs.py` and generated `E:/Projects/weksa/examples/speech-training/tiny-panphon-v0.1/batch.json` with fixed 1024/32/64 input widths, inline PanPhon 22-feature frames, and pending AquaSynth-owned 256/64 learned outputs."}
 {"date":"2026-05-24","topic":"cultmesh-speech-worker-authority","lesson":"CultMesh should distribute speech render/scoring work, not optimizer authority. AquaSynth emits typed SpeechRenderRequest documents; CultMesh delivers compiled Faust renderers, scoring code, small model/encoder assemblies, and batch/artifact refs to compatible leased workers; workers return SpeechRenderResult documents with loss, gradients, artifacts, and timing receipts. The authority trainer or shard-primary applies gradients through SpeechBackpropagationPipeline and commits SpeechTrainingCheckpoint documents through normal CultCache/CultMesh shard authority.","evidence":"Recorded the authority map in docs/ipa-vocal-tract-roadmap.md, state/spine.yaml, state/map.yaml, state/memory.json, and state/scratch.md. The next implementation layer is transport, admission, worker leases, payload delivery, and result collection; federated averaging waits until the simple authority path is boring."}
 {"date":"2026-05-24","topic":"cultmesh-speech-training-step","lesson":"AquaSynth can now run the first honest CultMesh-distributed speech training step: an authority trainer emits SpeechRenderRequest documents and a SpeechWorkerPayloadManifest, admits only CultMesh peers with valid speech-render-worker leases, records SpeechWorkerAdmissionReceipt documents for admitted/rejected peers, lets admitted workers produce SpeechRenderResult gradients, and applies those gradients centrally through SpeechBackpropagationPipeline. Workers still do not write checkpoints.","evidence":"Added GameCult.Mesh to AquaSynth.Core, added payload/admission/assignment/training-step types and store helpers in SpeechDistributedTraining.cs, added RunCultMeshTrainingStep, and added a test with two leased workers plus rejected unleased/read-replica peers. Focused verification: `dotnet test tests/AquaSynth.Dsl.Tests/AquaSynth.Dsl.Tests.csproj --filter \"SpeechDistributedTrainingTests\" -v minimal`: 3 passed. Remaining infrastructure gap: real remote artifact chunk transfer and process supervision belong to CultMesh, not an AquaSynth-local workaround."}
-{"date":"2026-05-25","topic":"vocal-tract-control-playground","lesson":"The current AquaSynth vocal-tract lane is touchable as a control surface before it is a Pink Trombone-class tract simulation. A useful playground should expose the real VocalTractControlTarget knobs directly, make the tract-ish shape visible, and stay honest that the WebAudio renderer is a lab proxy rather than full waveguide anatomy.","evidence":"Added tools/vocal-tract-playground with static HTML/CSS/JS, a Node static server, WebAudio source/filter synthesis, tract canvas visualization, presets, and sliders for the 14 base controls plus six mel bands. Verified with `node --check` on app.js/serve.mjs, curl against http://localhost:5125/, and browser interaction: Play toggled audio, Ee preset updated controls and canvas, console errors were empty."}
+{"date":"2026-05-25","topic":"vocal-tract-control-playground","lesson":"The current AquaSynth vocal-tract lane is touchable as a voice-patch target surface before it is a Pink Trombone-class tract simulation. A useful playground should expose the real VocalTractControlTarget knobs directly: these are the controls the learned speech driver predicts from phonetic and utterance input, not learned controls themselves. Keep the tract-ish visualization honest that the WebAudio renderer is a lab proxy rather than full waveguide anatomy.","evidence":"Added tools/vocal-tract-playground with static HTML/CSS/JS, a Node static server, WebAudio source/filter synthesis, tract canvas visualization, presets, and sliders for the 14 base voice-patch target controls plus six mel bands. Verified with `node --check` on app.js/serve.mjs, curl against http://localhost:5125/, and browser interaction: Play toggled audio, Ee preset updated controls and canvas, console errors were empty."}

state/scratch.md

@@ -4,8 +4,9 @@ Current slice:
 
 - Added `tools/vocal-tract-playground`, a static WebAudio playground for the
   current AquaSynth `VocalTractControlTarget` surface. It exposes the 14 base
-  speech controls plus six mel-envelope bands, draws a tract-ish canvas, and
-  includes vowel/noise/nasal presets.
+  voice-patch target controls plus six mel-envelope bands: these are the knobs
+  the learned speech driver will predict from phonetic and utterance input.
+  The page draws a tract-ish canvas and includes vowel/noise/nasal presets.
 - This is intentionally a touchable control playground, not a claim of
   Pink Trombone-class tract physics. The canvas and audio are tied to the real
   AquaSynth control names so future renderer work has an inspectable handle.

tools/vocal-tract-playground/README.md

@@ -14,7 +14,9 @@ Then open:
 http://localhost:5125/
 ```
 
-This is a control-surface witness, not a Pink Trombone-class tract simulation.
-It exposes the real AquaSynth speech-driver controls, draws a tract-shaped
-visualization, and uses a small WebAudio source/filter proxy so the knobs are
-audible while the proper tract renderer continues to mature.
+This is a voice-patch control witness, not a Pink Trombone-class tract
+simulation. It exposes the knobs the learned speech driver is meant to control:
+phonetic and utterance input go into the model, the model predicts this
+`VocalTractControlTarget` vector, and the voice patch renders sound from those
+values. The WebAudio source/filter proxy only makes that target vector touchable
+while the proper tract renderer continues to mature.

tools/vocal-tract-playground/app.js

@@ -387,10 +387,10 @@ function drawSpectralBars(width, height) {
 function drawReadout(width, height) {
   ctx.fillStyle = "#f0f4ed";
   ctx.font = "700 22px system-ui, sans-serif";
-  ctx.fillText(`vector[20] pressure=${state.pressure.toFixed(2)} noise=${state.turbulence.toFixed(2)} velum=${state.velum.toFixed(2)}`, 28, 42);
+  ctx.fillText(`voice patch target vector[20] pressure=${state.pressure.toFixed(2)} noise=${state.turbulence.toFixed(2)} velum=${state.velum.toFixed(2)}`, 28, 42);
   ctx.fillStyle = "#a8b3a3";
   ctx.font = "15px system-ui, sans-serif";
-  ctx.fillText("AquaSynth control target, not a claimed full tract simulation", 28, height - 24);
+  ctx.fillText("These are the knobs the learned speech driver will predict", 28, height - 24);
 }
 
 document.querySelectorAll("[data-preset]").forEach(button => {