Discrete sequence transformation primitives for MayaFlux::Kinesis. More...
This graph shows which files directly or indirectly include this file:
Go to the source code of this file.
Namespaces | |
| namespace | MayaFlux |
| Main namespace for the Maya Flux audio engine. | |
| namespace | MayaFlux::Kinesis |
| namespace | MayaFlux::Kinesis::Discrete |
Functions | |
| void | MayaFlux::Kinesis::Discrete::linear (std::span< double > data, double a, double b) noexcept |
| Linear map y = a*x + b applied in-place. | |
| void | MayaFlux::Kinesis::Discrete::power (std::span< double > data, double exponent) noexcept |
| Power map y = x^exponent applied in-place Scalar transcendental — not SIMD hot-path. | |
| void | MayaFlux::Kinesis::Discrete::exponential (std::span< double > data, double a, double b, double base=std::numbers::e) noexcept |
| Exponential map y = a * base^(b*x) applied in-place Scalar transcendental — not SIMD hot-path. | |
| void | MayaFlux::Kinesis::Discrete::logarithmic (std::span< double > data, double a, double b, double c, double base=std::numbers::e) noexcept |
| Logarithmic map y = a * log_base(b*x + c) applied in-place Values where (b*x + c) <= 0 are mapped to 0. | |
| template<typename TrigFunc > requires std::invocable<TrigFunc, double> && std::same_as<std::invoke_result_t<TrigFunc, double>, double> | |
| void | MayaFlux::Kinesis::Discrete::apply_trig (std::span< double > data, TrigFunc func, double frequency, double amplitude, double phase) noexcept |
| Trigonometric map y = amplitude * f(frequency*x + phase) applied in-place. | |
| void | MayaFlux::Kinesis::Discrete::clamp (std::span< double > data, double lo, double hi) noexcept |
| Clamp values to [lo, hi] in-place. | |
| void | MayaFlux::Kinesis::Discrete::quantize (std::span< double > data, uint8_t bits) noexcept |
| Quantize to n-bit resolution in-place Input is assumed to be in [-1, 1]; values outside are clamped first. | |
| void | MayaFlux::Kinesis::Discrete::normalize (std::span< double > data, double target_min=-1.0, double target_max=1.0) noexcept |
| Normalize to [target_min, target_max] in-place Single-pass min/max reduction followed by a single transform pass. | |
| void | MayaFlux::Kinesis::Discrete::reverse (std::span< double > data) noexcept |
| Reverse temporal order in-place. | |
| void | MayaFlux::Kinesis::Discrete::fade (std::span< double > data, double fade_in_ratio, double fade_out_ratio) noexcept |
| Apply equal-power (cosine) fade-in then fade-out envelope in-place The cosine taper maintains constant perceived loudness at the crossover point (0 dB centre gain vs −6 dB for a linear taper). | |
| std::vector< double > | MayaFlux::Kinesis::Discrete::slice (std::span< const double > data, double start_ratio, double end_ratio) |
| Extract a contiguous slice by ratio, returning a new vector. | |
| std::vector< double > | MayaFlux::Kinesis::Discrete::delay (std::span< const double > data, uint32_t delay_samples, double fill_value=0.0) |
| Prepend delay_samples zero-valued (or fill_value) samples, returning a new vector. | |
| void | MayaFlux::Kinesis::Discrete::interpolate_linear (std::span< const double > src, std::span< double > dst) noexcept |
| Linear interpolation from src into dst (caller sizes dst) Branchless inner loop with precomputed step; dst may be larger or smaller than src. | |
| void | MayaFlux::Kinesis::Discrete::interpolate_cubic (std::span< const double > src, std::span< double > dst) noexcept |
| Catmull-Rom cubic interpolation from src into dst (caller sizes dst) Branchless boundary clamping; Horner evaluation form. | |
| std::vector< double > | MayaFlux::Kinesis::Discrete::time_stretch (std::span< const double > data, double stretch_factor) |
| Time-stretch via linear interpolation resampling Fast but alias-naive: no anti-aliasing pre-filter is applied when stretch_factor < 1. | |
Detailed Description
Discrete sequence transformation primitives for MayaFlux::Kinesis.
Pure numerical functions operating on contiguous double-precision spans. No MayaFlux type dependencies. Domain-agnostic — the same primitives serve audio, visual, control, and any other sampled sequence.
All mutating functions operate in-place on the supplied span. Functions that change the output size return a new vector. Callers loop over channels; no multichannel wrappers are provided.
SIMD notes: linear, clamp, reverse, fade, normalize, quantize, interpolate_linear, and interpolate_cubic are written to be auto-vectorisable under -O2 -march=native (GCC/Clang). Transcendental functions (power, exponential, logarithmic) remain scalar; they are not hot-path and require SVML or a polynomial approximation for SIMD throughput.
Definition in file Transform.hpp.
