Complex waveforms: calculated or samples

[jupiter8]

thx, as said, I find this al very interesting. I am not a big fan of U-He's approach, but that is just my personal opinion.

Which one? He does all of them.

Modeling every resistor, diode, capacitor, etc.

That's just Diva. He does it in other ways in other synths so there's that. And a mix in some.
I'd say modelling is the "right way" to do it but that's just like my opinion man. It does come with a price (CPU consumption).

[Examigan]

The resynthesis in Icarus 3 is interesting and works well. I am not an expert on it, but maybe Markus might chime in?

[jupiter8]

Resynthesis are typically done with additive synthesis. Though i'd argue that it is the same thing as wavetable synthesis. Same coin different sides. I guess i'm a DSP nihilist "It doesn't matter, it's just a bunch of numbers"

With that said, Marcus seems to have thrown all the things at Icarus 3.

Classic wavetables, sample playback, FM, additive, PWM with any waveform, hypersaw, granular, looped waveforms, phase distortion, formant, waveshaping, ring modulation, amplitude modulation, oscillator sync, resonance, noise, lofi, 3D Wavetables, and many other modes exclusive to Icarus. All synthesis modes can be combined.

[Teksonik]

I am just interested in

You know what interests me? Using synths to make music. Nerding out on the specs and examining them with a scanning electron microscope seems like an utter waste of time to me.

I call that the Gentle Clock Divider syndrome. (those who know, know).

So to that end I'm going to stop talking about synths and go use mine for the joy of playing and composing........

[IvyBirds]

Modeled or sampled doesn't matter much for single cycle waveforms until you go to do something else with them

If you have calculated modeled waveforms you can start with a saw wave and have it seamlessly morph into a triangle wave

This what Omnisphere 3 does it uses DSP to model hundreds of potential wave forms including waveforms from many vintage Synths and let's you morph between them

[Korg Supporter]

Resynthesis are typically done with additive synthesis. Though i'd argue that it is the same thing as wavetable synthesis. Same coin different sides. I guess i'm a DSP nihilist "It doesn't matter, it's just a bunch of numbers"

With that said, Marcus seems to have thrown all the things at Icarus 3.

Classic wavetables, sample playback, FM, additive, PWM with any waveform, hypersaw, granular, looped waveforms, phase distortion, formant, waveshaping, ring modulation, amplitude modulation, oscillator sync, resonance, noise, lofi, 3D Wavetables, and many other modes exclusive to Icarus. All synthesis modes can be combined.

I thought wavetables were sample-based, but they can be edited using additive synthesis

[Essent]

With additive synthesis every waveform is calculated using sines (or something which resembles a sine). This way you can even calculate a saw too, however that is very computation intensive. So I was wondering how all these synths do it for these non standard waveforms.

[jupiter8]

Resynthesis are typically done with additive synthesis. Though i'd argue that it is the same thing as wavetable synthesis. Same coin different sides. I guess i'm a DSP nihilist "It doesn't matter, it's just a bunch of numbers"

With that said, Marcus seems to have thrown all the things at Icarus 3.

Classic wavetables, sample playback, FM, additive, PWM with any waveform, hypersaw, granular, looped waveforms, phase distortion, formant, waveshaping, ring modulation, amplitude modulation, oscillator sync, resonance, noise, lofi, 3D Wavetables, and many other modes exclusive to Icarus. All synthesis modes can be combined.

I thought wavetables were sample-based, but they can be edited using additive synthesis

Yeah you're not wrong. Conceptually it is basically the same thing though. Just a different way of storing the waveforms.

It even mentions wavetables as an efficient way of implementing additive synthesis on Wikipedia i just noticed.

[IvyBirds]

With additive synthesis every waveform is calculated using sines (or something which resembles a sine). This way you can even calculate a saw too, however that is very computation intensive. So I was wondering how all these synths do it for these non standard waveforms.

Additive synthesis works because in the real world limitless sine waves of various frequencies combine together into one "resultant" wave. That wave then hits your ear drum and causes it to vibrate that vibration causes electrical impulses to be generated which you brain interprets as a specific sound

So you can just start adding sine waves of various frequencies and amplitudes and you can make your own "resultant" wave. In a musical context those frequencies are based around a fundamental frequency and harmonics above it so we get stable pitches.

With PCM sampling these resultant waves created naturally or via synthesis are run though a microphone which turns them into a voltage which is then run through an Analog-to-Digital Converter that measure their value thousands of times a second and stores those values as digital code

Then at playback the digital code is run through a Digital to Analog Converter which converts back to a voltage which can make a speaker vibrate to get that resultant wave. Depending on the bit depth and sample rate of your PCM Sample the quality of the playback could be pristine or low fidelity and full of noise and errors

With synths that Digital Signal Processing to generate and calculate waveforms they just bypass the actual sampling stage and use a computer to generate the math which sends code to the DAC to make the resultant waves

The end you can use Additive, sampling, or DSP. The quality of the code that goes to the DAC might be the same or it might be different and the end result is actually sound you hear might be slightly different

[IvyBirds]

Resynthesis are typically done with additive synthesis. Though i'd argue that it is the same thing as wavetable synthesis. Same coin different sides. I guess i'm a DSP nihilist "It doesn't matter, it's just a bunch of numbers"

With that said, Marcus seems to have thrown all the things at Icarus 3.

Classic wavetables, sample playback, FM, additive, PWM with any waveform, hypersaw, granular, looped waveforms, phase distortion, formant, waveshaping, ring modulation, amplitude modulation, oscillator sync, resonance, noise, lofi, 3D Wavetables, and many other modes exclusive to Icarus. All synthesis modes can be combined.

I thought wavetables were sample-based, but they can be edited using additive synthesis

Historically Wavetables were sample based.

Modern Computers are cheap and powerful enough that they can use DSP to generate the waves on demand for each Wavetable. This is a serious advantage because it can also generate values in-between the individual waves this allows those waves to be morphed. So you can go from a square wave to a triangle wave smoothly and slowly morph between the two

With samples you can't do that

However single cycle waveforms in a sample based Wavetable can be broken down into the individual sine waves that make up that sound using a mathematical formula and those sine wave values can be loaded into an additive synth engine. Once it's in an additive engine you can use that to morph between say a square and a triangle

[Korg Supporter]

Resynthesis are typically done with additive synthesis. Though i'd argue that it is the same thing as wavetable synthesis. Same coin different sides. I guess i'm a DSP nihilist "It doesn't matter, it's just a bunch of numbers"

With that said, Marcus seems to have thrown all the things at Icarus 3.

Classic wavetables, sample playback, FM, additive, PWM with any waveform, hypersaw, granular, looped waveforms, phase distortion, formant, waveshaping, ring modulation, amplitude modulation, oscillator sync, resonance, noise, lofi, 3D Wavetables, and many other modes exclusive to Icarus. All synthesis modes can be combined.

I thought wavetables were sample-based, but they can be edited using additive synthesis

Historically Wavetables were sample based.

Modern Computers are cheap and powerful enough that they can use DSP to generate the waves on demand for each Wavetable. This is a serious advantage because it can also generate values in-between the individual waves this allows those waves to be morphed. So you can go from a square wave to a triangle wave smoothly and slowly morph between the two

With samples you can't do that

However single cycle waveforms in a sample based Wavetable can be broken down into the individual sine waves that make up that sound using a mathematical formula and those sine wave values can be loaded into an additive synth engine. Once it's in an additive engine you can use that to morph between say a square and a triangle

Hive has a couple of interpolation options. One is spectral, and the other is more conserved

[IvyBirds]

Hive has a couple of interpolation options. One is spectral, and the other is more conserved

Hive has four plugin interpolation algorithms options to determine how it transitions between frames, allowing for a balance between audio quality and CPU usage.

The first is Switch where no interpolation occurs; the oscillator switches directly between frames, this can create stepping which may or may not sound good

The second is Crossfade which smoothly blends linear changes between wavetable frames.

Both of those are very CPU efficient and are not generating any new waves

The third and fourth are called "Spectral" and "Zero Phase"

Both of those use additive synthesis.

Spectral is using additive synthesis to make new waveforms between the frames of the Wavetable. It also tries to save the character of each frame however by focusing on the magnitude and the phase of them. This can be CPU intensive and can lead to come filtering and phase cancelations of the individual harmonics in the additive synthesis

The Zero Phase Algorithm prevents phase cancelations by also focusing on the phase of each partial in additive synthesis and making each of them start at 0 phase. This comes at a cost however with higher CPU usage and also potential loss of character of the underlying frames in the wave table