Wavetable support for OneCycleWaveform

[ryukau]

This is a requested feature from a user.

What to do:

• Join several renderings to one file.
• Randomize each rendering.

There's a synth called Wavetable in Live. It can take a wave file with length of 1024 * n samples. I'm not sure if this is accurate, but it seems like each 1024 sample block is assigned to a wavetable frame.

Preparing test environment is the current blocker. I don't have wavetable synth that can read 2 or more frames of waveforms. I'd like to test the result in some way.

For randomization, maybe add something like random amount, and use that value for every randomization. The scaling for each parameter becomes a problem. Because it takes time to tune them all.

[NeuroNoNeuro]

Possibly offer an option of:

1. "Create Wavetable"
2. Select quantity of 2 to 100 OneCycleWaveform renderings
3. Press the "Random" button
4. Output a single audio file with 100 OneCycleWaveforms 'glued' together

If possible, the arrangement of renderings could go from soft (sine wave) to harsh (noise) in order. Like so:

This output would be ideal, but even a randomized wavetable would be exciting, also creating much less work, not having to organize the arrangement of waveform timbre.

[ryukau]

Sorting spectra is not so straight forward. But in this case, something like spectral whiteness can be used.

[ryukau]

Random multiple table generation is added to OneCycleWaveform.

• nTable is the number of wavetables.
• It will randomize only when Random Amount is not 0.

@NeuroNoNeuro Let me know if it works on Wavetable.

[NeuroNoNeuro]

Oh WOW!!! I was just going to build some more wavetables by hand, and now this is here. Saves me a literal TON of time. Works great! I just messed with it a little, and chose to manipulate only:

N-Table: 10
Seed: 0
Amount: 25

Works great! Holy cow dude. Exported the file, dropped it in and BOOM. Done. Perfect.

[ryukau]

I'm glad it worked! 😄

For the record, sorting is somewhat arbitrary. There's a opportunity for different musical tunings.

[NeuroNoNeuro]

One question - how would one go about creating additive-type waveforms - I know it can, because I get some through random generation. What would be a way to get a rudimentary set of additive tones - from sine wave up to varying sorts of sine-waves with odd pulses?

[ryukau]

Could you provide an example output? However, I probably have no answer for that. Maybe Spectral and Filter sections are related to it.

One problem is that the sorting is flaky, that the result doesn't always sort from sine (less overtones) to pulse/glitch (more overtones). So it's hard to expect the result. Other is the parameter tunings on randomization. For this, I need testing environment but I don't have one. Below is the link to the relevant code.

UhhyouWebSynthesizers/OneCycleWaveform/renderer.js

Lines 64 to 93 in 86ae178

	const tiltLin = (rng, base, lower, upper, range) => {
	const value = base + uniformDistributionMap(rng.number(), -range, range);
	return clamp(value, lower, upper);
	};
	const tiltExp = (rng, base, lower, upper, range) => {
	const logRange = range * Math.log(upper / lower);
	const low = Math.max(lower, base * Math.exp(-logRange));
	const high = Math.min(upper, base * Math.exp(logRange));
	return base * exponentialMap(rng.number(), low, high);
	};
	const waveform = tiltLin(rng, pv.waveform, 0, 1, pv.randomAmount);
	const powerOf = tiltExp(rng, pv.powerOf, 0.01, 100, pv.randomAmount);
	const skew = tiltExp(rng, pv.skew, 0.01, 100, pv.randomAmount);
	const sineShaper = tiltLin(rng, pv.sineShaper, 0, 1, pv.randomAmount);
	const sineRatio = Math.floor(tiltExp(rng, pv.sineRatio, 1, 1024, pv.randomAmount));
	const hardSync = tiltExp(rng, pv.hardSync, 0.1, 10, pv.randomAmount);
	const mirrorRange = tiltLin(rng, pv.mirrorRange, 0, 1, pv.randomAmount);
	const mirrorRepeat = tiltLin(rng, pv.mirrorRepeat, 0, 1, pv.randomAmount);
	const flip = tiltLin(rng, pv.flip, -1, 1, pv.randomAmount);
	const spectralSpread = tiltExp(rng, pv.spectralSpread, 0.01, 100, pv.randomAmount);
	const phaseSlope = tiltExp(rng, pv.phaseSlope, 0.001, 1000, pv.randomAmount);
	const highpass = tiltExp(rng, pv.highpass, 0.001, 1, pv.randomAmount);
	const lowpass = tiltExp(rng, pv.lowpass, 0.001, 1, pv.randomAmount);
	const notchStart = tiltExp(rng, pv.notchStart, 0.001, 1, pv.randomAmount);
	const notchRange = tiltExp(rng, pv.notchRange, 0.001, 1, pv.randomAmount);
	const lowshelfEnd = tiltExp(rng, pv.lowshelfEnd, 0.001, 1, pv.randomAmount);
	const lowshelfGain = tiltExp(rng, pv.lowshelfGain, 0.01, 100, pv.randomAmount);

[NeuroNoNeuro]

Unfortunately I cannot. I can say that your PADchoir synth can create the waveforms if the user selects ‘additive pad’.