FnSignal.kt

src/jvmMain/kotlin/afield/audio/synth/FnSignal.kt · 40 lines

@file:afield.runtime.completion.CompletionCategory("Synthesis")
package afield.audio.synth

import afield.math.functions.Functions

Plays a Functions.Fn of time as audio, evaluating it at every sample rather than once per block.

FnSignal wraps the function and toSignal is the extension that builds one, so an envelope or frequency contour written as an Fn can drive a synth graph sample-accurately.

class FnSignal(val fn: Functions.Fn) : Signal()

A Functions.Fn of time, in seconds, heard as a Signal — evaluated at every sample.

Sample i of a block is fn(time + i / sampleRate), where time is SynthContext.time, the start of the block. So a function used as an amplitude envelope or a frequency contour reaches the audio at the instant it names, with no hold, no ramp and no delay.

That is the difference from the other way to bring a function into a graph, Param { fn(time) }. A Param supplier is evaluated once per block, at the block's start, and Param ramps from the previous block's value to this one across the block — so what the audio hears at t is, to first order, the function at t minus one block. At Dough's 128-frame block and 48 kHz that is 2.67 ms. chain_fn_sound_roundtrip.md measures both routes from the rendered .wav.

The cost is one Functions.Fn.eval per sample, and an Fn evaluates its whole graph each time, so a deep graph is expensive at audio rate. Values are narrowed to Float on the way out. The function's domain is not consulted — like Functions.Fn.eval, a time outside it is passed through.

override fun generate(output: FloatArray, numFrames: Int, ctx: SynthContext)

    override fun generate(output: FloatArray, numFrames: Int, ctx: SynthContext) {
        val t0 = ctx.time.toDouble()
        val dt = 1.0 / ctx.sampleRate.toDouble()
        for (i in 0 until numFrames) output[i] = fn.eval(t0 + i * dt).toFloat()
    }
}

fun Functions.Fn.toSignal(): Signal = FnSignal(this)

This function of time, in seconds, as a Signal evaluated at every sample — see FnSignal.