GameCult/AquaSynth
Watch
9
0
Fork
You've already forked AquaSynth
 0
Code Issues Pull requests 4 Projects Releases Packages Wiki Activity Actions
AquaSynth/docs/zynaddsubfx-pressure-survey.md
Meta Vens 2461821907 Add Zyn formant parity fixtures
2026-05-17 20:52:34 +01:00

9.9 KiB
Raw Blame History

ZynAddSubFX Pressure Survey

Objective

Find the ZynAddSubFX instruments most likely to break AquaSynth's current patch language before spending time on syntax polish.

This is a feature-pressure survey, not an audio parity claim. The upstream Zyn instrument bank is GPL reference feedstock through the pinned Zyn submodule tree. It is development/test material only: useful for reports, parity renders, and pressure selection, never package content or AquaSynth patch-library stock.

Source Checked

  • Repository: https://github.com/zynaddsubfx/instruments
  • Observed commit: c5c912131b31df5fdf372d2f06a25aaf2375837f
  • Files scanned: 1318 .xiz instruments
  • Reader result after hardening: 1318 parsed, 0 errors
  • Local path: external/zynaddsubfx/instruments/banks

Scoring

ZynInstrumentSurvey.ComplexityScore ranks instruments by enabled kit items, layering, active ADD/SUB/PAD engines, envelope count, free-envelope count, LFO count, filter count, formant-filter count, and effect count.

The score is deliberately crude. It is a triage knife, not a musicological thesis.

Top Pressure Targets

Score Path Name Why It Hurts
666 olivers-100/0032-Drum Kit.xiz Drum Kit 16 enabled kit items, ADD/SUB/PAD mixed engines, 106 envelopes, 25 free envelopes, 73 LFOs, 37 filters, 3 formant filters. This is the named-layer/kit-routing boss fight.
398 Companion/0056-Wide Bass.xiz Wide Bass Only 2 enabled kit items, but huge modulation/formant pressure: 86 envelopes, 54 LFOs, 18 filters, 18 formant filters. This is vowel/formant motion pressure wearing bass clothes.
357 Cris Owl Alvarez/0160-DRUMKIT toy drummer 2.xiz DRUMKIT toy drummer 2 12 enabled ADD kit items with 77 envelopes, 9 free envelopes, 38 LFOs, and 32 filters. Layered percussion without PAD escape hatches.
260 olivers-other/0038-Deep Zyn.xiz Deep Zyn 4 ADD layers and 36 free envelopes. This is the free-envelope case without relying on mixed engines.
251 Companion/0121-Ghost Ensemble.xiz Ghost Ensemble ADD/SUB/PAD all active, layered, 50 envelopes, 30 LFOs, 11 filters, 11 formant filters. Good broad-spectrum stress target.
220 Drums/0001-Drums Kit1.xiz Drums Kit 11 enabled kit items, ADD/SUB mix, 30 envelopes, 11 free envelopes, 15 LFOs, 12 filters. A smaller drum-kit target than Oliver's monster.
181 olivers-100/0124-Simple Clonewheel.xiz Simple Clonewheel 8 enabled kit items with 7 PAD engines. Best early pressure for layered PAD routing without drum semantics.

Upstream PAD Listening Batch

ZynPadReferenceRendererSurveysUpstreamGplPadFixturesWhenBuilt renders a small GPL PAD batch through the real Zyn oracle when the renderer is built:

  • Pads/0002-sin2x pad.xiz
  • Pads/0065-Soft Pad.xiz
  • Dual/0008-Organ Choir Pad2.xiz
  • Laba170bank/0098-DoublePadBass.xiz
  • Companion/0121-Ghost Ensemble.xiz

Each render lands in a per-synth timestamped folder under artifacts/parity/zyn-upstream-pad-fixtures/, with Zyn reference WAVs, source table WAVs, report.txt, and candidate.aqua. The candidate is now generated per preset by ZynInstrumentReader.RebuildFirstPadAsAquaSynthScript: first enabled PAD kit item, OSCIL harmonic magnitudes, oracle-reported table basefreq, and basic volume/envelope scaffolding. The report names missing semantics such as Zyn harmonic profile shaping, bandwidth/profile behavior, and filter envelope/LFO lowering. These are listening and parity fixtures, not library candidates.

Upstream Formant Listening Batch

ZynVoiceReferenceRendererSurveysUpstreamFormantFixturesWhenBuilt renders a small GPL ADD/formant batch through ZynVoiceReference when the reference renderer is built:

  • Companion/0056-Wide Bass.xiz
  • Companion/0121-Ghost Ensemble.xiz
  • Companion/0119-Ghost Choir.xiz

The batch intentionally requires multi-frame formant motion so the first workout does not get hijacked by drum-kit envelope/source failures. Each run writes Zyn reference WAVs, AquaSynth candidate WAVs, candidate.aqua, and report.txt under artifacts/parity/zyn-upstream-formant-fixtures/. The first baseline scores are poor but useful: Wide Bass 0.280990, Ghost Ensemble 0.312813, and Ghost Choir 0.363644. The shared missing features are exact Zyn filter-position-driven vowel motion, formant smoothing/clearness, and ADD voice source detail.

This is now ready for the real parity workout. The fixtures are expected to scream. The point is that they scream in a chorus instead of letting one wounded patch write the architecture.

Pressure Conclusions

  • Named layers are real. Anonymous AquaSynth voices can sketch small Zyn layers, but drum kits and clonewheel-style instruments need stable layer names, key ranges, engine identity, and per-layer routing.
  • Free envelopes are not optional forever. Zyn's worst cases use dozens of free envelopes. ADSR plus staged operator envelopes will not remain honest for normal voice-layer work.
  • Formant motion is bigger than static formant filters. Wide Bass and Ghost Ensemble show that static formant banks are only the foothold.
  • PAD is both source and layer pressure. PADsynth-style patches are not just saw pads with slow attack; they need a spectral/wavetable source authority.
  • Drum kits should not be first golf targets. They are the best stress test for the model, but probably too broad for the first new syntax slice.

Coverage Map

ZynInstrumentSurvey.CoverageDirectory now emits a handled/counted/unknown coverage report instead of only ranking instruments by aggregate pressure. The upstream GPL bank report is written to artifacts/parity/zyn-upstream-bank-survey/coverage.md by SurveysUpstreamGplInstrumentBankWhenAvailable.

Current upstream bank coverage:

  • Files parsed: 1318/1318
  • Enabled kit items: 1459
  • Layered instruments: 63
  • Handled with parity-backed or native AquaSynth coverage: native band-pass and notch filters, PAD layers, named kit layers, PAD table modes/bandwidth, PAD source-table extraction for the implemented OscilGen subset, PAD LP/HP/BPF/notch cutoff/order/Q/envelopes, staged non-free envelopes, and authored vowel-frame formant motion.
  • Counted but not translated: ADD, SUB, mixed-engine kits, free envelopes, active LFO routing beyond explicit control lanes, exact Zyn formant position drivers/smoothing, and effects.
  • Unknown PAD filter pressure: formant filters, peak filters, shelf filters, and other unimplemented PAD global filter categories/types such as cat=1 type=0, cat=0 type=6, cat=0 type=8, and smaller buckets. Analog and state-variable band-pass/notch (cat=0 type=4/5, cat=2 type=2/3) now lower through AquaSynth's native bpf/notch filter authority.
  • Unknown PAD OscilGen pressure: unimplemented source filter types 10, 12, 11, 4, 5, 3, and 8; unimplemented waveshapers 2, 14, 5, 6, 3, 13, 8, and 9.

This map is not a clone checklist. It is a workout list. A bucket graduates from counted or unknown only when a parity target proves which AquaSynth authority should own the behavior.

Recommended Next Cut

Start with ADD/SUB voice extraction or free-envelope point curves, not more PAD synthesis.

Reason: named layers, PAD spectral sources, staged envelopes, native band-pass/notch filters, and a first vowel-frame formant-motion surface now exist. The next counted Zyn pressure is no longer "where does this live?" for formants; it is either ADD/SUB source detail or arbitrary envelope shape.

First Responses

AquaSynth now has a minimal layer command:

layer name=pad_low engine=pad min_key=36 max_key=84 gain=.07
voice layer=pad_low freq=130.8128

The layer owns source identity and metadata while current lowering still emits ordinary voices. This is the scaffold needed before promoting PAD sources, additive harmonic banks, free envelopes, or richer formant motion.

Additive layers now also have a harmonics command:

layer name=body engine=add gain=.16
harmonics layer=body root=220 partials=1:.16,2:.075,3:.045

The harmonic bank records authored additive partial intent under a named layer, then lowers to ordinary sine voices. It does not claim Zyn ADDsynth's full oscillator table: phase, bandwidth, oscillator shaping, and exact free-envelope behavior remain separate pressure.

Normal voices can also use staged rate/level envelopes through the same env=rl surface already used by operator graphs:

layer name=pad engine=pad env=rl rates=.5,.7,1.0,1.4 levels=1,.85,.7,0 gate=2.2
voice layer=pad freq=220

That gives PAD and vocal layers asymmetric contour authority without returning to anonymous ADSR piles. Arbitrary Zyn free-envelope point curves remain pressure; the staged surface is the first coherent rung, not the whole ladder.

PAD layers now have a spectral-cloud source command:

layer name=pad_low engine=pad
spectrum layer=pad_low root=130.8128 spread=.012 partials=1:.07,1.5:.052

The command records PAD-like partial intent under the layer and emits a static Faust wavetable source before normal voice treatment. The table is generated by a PADsynth-style authoring path: frequency-domain harmonic spreading, deterministic random phase, one inverse FFT, and normalization. This is not full Zyn PAD engine parity across every harmonic profile and pitch-zone behavior, but it is now real spectral-table synthesis rather than voice-count theater.

Wide Bass promoted the first formant-motion rung:

voice layer=wide_bass formant_mix=.7 vowel_hz=.39
    vowels=600:90:1,1200:160:.7|500:80:1,900:120:.8

The vowels field is an AquaSynth voice authority: each |-separated frame is a formant bank, and Faust blends adjacent frames cyclically at vowel_hz. The Zyn workout can seed frames from FORMANT_FILTER vowel sequences, but exact Zyn filter-frequency position driving, slowness, and vowel clearness remain pressure rather than hidden compatibility sludge.