- Generated by
1.15.0
The calibration register: every number this module ships that is not cited to the literature lives here with a default and a range, so a reviewer finds them in one place rather than spread through the code. More...
Classes | |
| struct | VelvetTapDesign |
| Pure intermediate for one candidate in a velvet tap grid. More... | |
Functions | |
| constexpr double | allpass_l1_gain (double g) |
| The time-domain L1 (peak) gain of one Schroeder allpass with coefficient g: Σ|h| = g + (1−g²)/(1−g) = 1 + 2g. | |
| double | worst_case_gain (double diffusion=kDiffusionDefault, bool converter_on=false) |
| The shipped worst-case gain as a closed form of the shipped constants: Π_i (1 + 2·g_i) · G_L1, an upper bound on Σ_n |h[n]| of the rendered impulse response by Young's convolution inequality. | |
| double | program_envelope (NonlinProgram program, double tau, double gate_hold=kGateHold, double attack=kRevRise) |
| The dimensionless program envelope E(τ), τ ∈ [0, 1] (series law 7 — scale invariant, so changing length_ms rescales time without refitting anything). | |
| double | pulse_density (double u, double density_pct, double gamma) |
| Instantaneous pulse density in pulses/second at normalized time u ∈ [0,1]: Nd(u) = Nd_min + (Nd_max − Nd_min)·u^γ. | |
| VelvetTapDesign | design_velvet_tap (NonlinProgram program, double position_samples, double grid_samples, int window_samples, int predelay_samples, double gate_hold, double attack, const VelvetNoiseDrawT< double > &draw) noexcept |
Variables | |
| constexpr int | kNumAllpass = 2 |
| Number of pre-diffusion Schroeder allpasses. | |
| constexpr double | kAllpassDelaysMs [2] = {4.77, 6.13} |
| Allpass delays in ms, rounded to the nearest sample and then nudged to the nearest prime sample count so the two combs never align. | |
| constexpr double | kDiffusionDefault = 0.70 |
| Allpass coefficient, exposed as diffusion. | |
| constexpr double | kDiffusionMax = 0.85 |
| constexpr double | kNdMin = 800.0 |
| Instantaneous pulse density at the start and end of the window, in pulses/second. | |
| constexpr double | kNdMax = 4000.0 |
| constexpr double | kDensityRefPct = 60.0 |
| The density_pct value at which Nd_min equals kNdMin exactly, so the shipped default reproduces the shipped constant rather than a scaled version of it. | |
| constexpr double | kGammaDefault = 2.0 |
| Density growth exponent γ. | |
| constexpr double | kAmbDropDb = 60.0 |
| Ambience drop across the window, in dB; sets α = A_drop·ln(10)/20. | |
| constexpr double | kFadeInFrac = 0.02 |
| Linear rise over the first fraction of the window, which kills the first-tap click on the programs that would otherwise start at full gain. | |
| constexpr double | kGateHold = 0.70 |
| Gated program: normalized hold point h and fall width w. | |
| constexpr double | kGateFall = 0.05 |
| constexpr double | kRevRise = 0.85 |
| Reverse program: normalized rise point r and swell exponent p (p < 1 concave, p > 1 convex). | |
| constexpr double | kRevPow = 1.0 |
| constexpr int | kNlHumps = 2 |
| NonLin2 program: number of raised-cosine humps, ripple depth, gate point. | |
| constexpr double | kNlDepth = 0.4 |
| constexpr double | kNlHold = 0.85 |
| constexpr double | kL1Budget = 4.0 |
| The per-channel L1 budget of the tap-gain vector. | |
| constexpr int | kSegments = 8 |
| Number of time segments the tap range is split into for spectral tilt. | |
| constexpr double | kFcBright = 16000.0 |
| Segment tilt corners. | |
| constexpr double | kFcDark = 6000.0 |
| constexpr double | kSwapFadeMs = 20.0 |
| Equal-power crossfade length applied when a topology change swaps tap banks. | |
| constexpr double | kParamSmoothMs = 20.0 |
| Smoothing time for the continuous gain parameters. | |
| constexpr std::uint32_t | kSeedRodd = 0x9E3779B9u |
| seed_R = seed_L ⊕ kSeedRodd. | |
| constexpr std::uint32_t | kDefaultSeed = 0x51EDu |
| Default seed_L. | |
| constexpr int | kTapGuard = 32 |
| Headroom above the expected tap count, so the jittered grid can never outgrow the pre-sized table. | |
| constexpr double | kConverterFcLo = 80.0 |
| Converter-character pre-filter corners (§5). | |
| constexpr double | kConverterFcHi = 15000.0 |
| constexpr int | kConverterBitsMax = 16 |
| Converter word length at converter_amount = 0 and the depth of the sweep: bits = kConverterBitsMax − round(kConverterBitSweep · amount). | |
| constexpr int | kConverterBitSweep = 4 |
| constexpr double | kConverterDcPole = 0.995 |
| DC-blocker pole used by the converter stage. | |
| constexpr double | kMinLengthMs = 20.0 |
| Largest window and pre-delay prepare() sizes storage for, in ms. | |
| constexpr double | kMaxLengthMs = 2000.0 |
| constexpr double | kMaxPredelayMs = 200.0 |
| constexpr double | kMinDensityPct = 10.0 |
| Lower bound of the density control's documented 10 .. 100 % range. | |
| constexpr double | kNormFloor = 1e-20 |
| Floor below which the normalization sum is treated as empty, so a program whose taps all land in a zeroed region produces silence rather than a division by a denormal. | |
| constexpr double | kDcBlockerL1Gain = 2.0 |
| The time-domain L1 gain of the DC blocker, which is exactly 2 for any pole p < 1: h[0] = 1, h[n] = −(1−p)p^{n−1}, so Σ|h| = 1 + (1−p)/(1−p). | |
The calibration register: every number this module ships that is not cited to the literature lives here with a default and a range, so a reviewer finds them in one place rather than spread through the code.
|
constexpr |
The time-domain L1 (peak) gain of one Schroeder allpass with coefficient g: Σ|h| = g + (1−g²)/(1−g) = 1 + 2g.
Exact, not an approximation.
|
inlinenoexcept |
|
inline |
The dimensionless program envelope E(τ), τ ∈ [0, 1] (series law 7 — scale invariant, so changing length_ms rescales time without refitting anything).
gate_hold is the Gated program's h; attack is the Reverse program's r and the NonLin2 program's h. Both are normalized fractions, not percents. Neither is read by Ambience.
|
inline |
Instantaneous pulse density in pulses/second at normalized time u ∈ [0,1]: Nd(u) = Nd_min + (Nd_max − Nd_min)·u^γ.
The u^γ law is the geometric- acoustics t² density growth (Moorer 1979) with γ exposed, because the gated and reverse programs deliberately break that physics.
|
inline |
The shipped worst-case gain as a closed form of the shipped constants: Π_i (1 + 2·g_i) · G_L1, an upper bound on Σ_n |h[n]| of the rendered impulse response by Young's convolution inequality.
converter_on adds the converter stage's DC blocker. This is the value the registry cites and the value the test suite asserts the rendered response stays under.
|
inlineconstexpr |
Allpass delays in ms, rounded to the nearest sample and then nudged to the nearest prime sample count so the two combs never align.
[design parameter] default {4.77, 6.13}, range 1 .. 30 ms each.
|
inlineconstexpr |
Ambience drop across the window, in dB; sets α = A_drop·ln(10)/20.
[design parameter] default 60 dB, range 20 .. 80 dB.
|
inlineconstexpr |
Converter word length at converter_amount = 0 and the depth of the sweep: bits = kConverterBitsMax − round(kConverterBitSweep · amount).
[design parameter] defaults 16 / 4, ranges 8 .. 24 / 0 .. 8.
|
inlineconstexpr |
|
inlineconstexpr |
DC-blocker pole used by the converter stage.
[design parameter] default 0.995 (≈3.5 Hz at 44.1 kHz), range 0.9 .. 0.9999.
|
inlineconstexpr |
|
inlineconstexpr |
Converter-character pre-filter corners (§5).
No verifiable public converter spec exists for any hardware, so these are ours. [design parameter] defaults 80 / 15000 Hz, ranges 20 .. 200 / 10k .. 20k Hz.
|
inlineconstexpr |
The time-domain L1 gain of the DC blocker, which is exactly 2 for any pole p < 1: h[0] = 1, h[n] = −(1−p)p^{n−1}, so Σ|h| = 1 + (1−p)/(1−p).
|
inlineconstexpr |
Default seed_L.
Seeds are a preset choice, never automated (series law 2). [design parameter] default 0x51ED, range 0 .. 2³¹−1.
|
inlineconstexpr |
The density_pct value at which Nd_min equals kNdMin exactly, so the shipped default reproduces the shipped constant rather than a scaled version of it.
[design parameter] default 60 %, range 10 .. 100 %.
|
inlineconstexpr |
Allpass coefficient, exposed as diffusion.
|g| < 1 guarantees stability. [design parameter] default 0.70, range 0 .. 0.85.
|
inlineconstexpr |
|
inlineconstexpr |
Linear rise over the first fraction of the window, which kills the first-tap click on the programs that would otherwise start at full gain.
[design parameter] default 0.02, range 0 .. 0.1.
|
inlineconstexpr |
Segment tilt corners.
kFcDark is the default of the hf_damp_hz parameter. [design parameter] defaults 16000 / 6000 Hz, ranges 4k .. 20k / 1k .. 18k.
|
inlineconstexpr |
|
inlineconstexpr |
Density growth exponent γ.
2.0 is the physical t² reflection-density growth of geometric acoustics (Moorer 1979); 0 is flat density, which is how the gated and reverse programs deliberately break physics. [design parameter] default 2.0, range 0 .. 2.
|
inlineconstexpr |
|
inlineconstexpr |
Gated program: normalized hold point h and fall width w.
[design parameter] defaults 0.70 / 0.05, ranges 0.1 .. 0.95 / 0.01 .. 0.2.
|
inlineconstexpr |
The per-channel L1 budget of the tap-gain vector.
This sets both the loudness and the peak-gain bound: Σ_k |g_k| = G_L1 exactly, by construction. [design parameter] default 4.0, range 1.0 .. 8.0.
|
inlineconstexpr |
|
inlineconstexpr |
|
inlineconstexpr |
Lower bound of the density control's documented 10 .. 100 % range.
|
inlineconstexpr |
Largest window and pre-delay prepare() sizes storage for, in ms.
These are the maxima of the corresponding parameter-table rows. [design parameter] defaults 2000 / 200 ms, ranges 100 .. 8000 / 0 .. 1000 ms.
|
inlineconstexpr |
|
inlineconstexpr |
Instantaneous pulse density at the start and end of the window, in pulses/second.
Velvet noise is subjectively smooth by ≈2000 pulses/s and fully smooth by ≈4000 (Karjalainen & Järveläinen 2007; Välimäki et al. 2017), so the upper value is what guarantees a smooth late field. [design parameter] defaults 800 / 4000, ranges 200 .. 2000 / 2000 .. 6000.
|
inlineconstexpr |
|
inlineconstexpr |
|
inlineconstexpr |
NonLin2 program: number of raised-cosine humps, ripple depth, gate point.
[design parameter] defaults 2 / 0.4 / 0.85, ranges 1 .. 4 / 0 .. 0.8 / 0.05 .. 0.98.
|
inlineconstexpr |
Floor below which the normalization sum is treated as empty, so a program whose taps all land in a zeroed region produces silence rather than a division by a denormal.
Guards a divide; any value far below the smallest meaningful L1 sum works. Range 1e-30 .. 1e-9.
|
inlineconstexpr |
Number of pre-diffusion Schroeder allpasses.
[design parameter] default 2, range 0 .. 4.
|
inlineconstexpr |
Smoothing time for the continuous gain parameters.
[design parameter] default 20 ms, range 1 .. 200 ms.
|
inlineconstexpr |
|
inlineconstexpr |
Reverse program: normalized rise point r and swell exponent p (p < 1 concave, p > 1 convex).
[design parameter] defaults 0.85 / 1.0, ranges 0.5 .. 0.98 / 0.5 .. 2.0.
|
inlineconstexpr |
seed_R = seed_L ⊕ kSeedRodd.
Honest gap: no citable literature exists for this exact 32-bit constant's adoption here. It is the conventional golden-ratio-derived (2³²/φ) multiplicative-hash odd constant, used only because it disperses bits well; any odd 32-bit constant with good dispersion would decorrelate the two seeds equally. A fixed algorithmic constant with a numerical rationale, not a musical design parameter — no range applies.
|
inlineconstexpr |
Number of time segments the tap range is split into for spectral tilt.
Per-tap filtering would be O(taps × filter); this is O(taps + S). [design parameter] default 8, range 4 .. 16.
|
inlineconstexpr |
Equal-power crossfade length applied when a topology change swaps tap banks.
[design parameter] default 20 ms, range 5 .. 100 ms.
|
inlineconstexpr |
Headroom above the expected tap count, so the jittered grid can never outgrow the pre-sized table.
[design parameter] default 32 taps, range 0 .. 128.
1.15.0