bazille fractal resonance osc = just a formant osc right ?

[gentleclockdivider]

So after a comparing bazille fractal reso. oscillators with a reaktor ensemble called goldvosim ( an old one from dashsynthesis )
I wonder why they just don't call it a formant oscillator ..;because that is what the fractalize knob is doing ..
Users of reaktor goldvosim will now what I mean

[EvilDragon]

Because "fractal resonance" just sounds infinitely more cooler!

[Urs]

Not sure. What we're doing is the same as what Casio did in the CZ series for resonance. Only, we use the waveform itself instead of just a sine. Fractal because it reminded me of fractals, Resonance because this is the origin of the idea.

No claim of novelty here (we're not you know who)

[Sendy]

A funny thing you can do in Bazille is have a resonant wave inside a resonant wave. It'll alias if you send one of the resonance frequencies up high, but all you're missing in this case is the resonance being too high to really hear, so it's not a big deal. You can have a resonant frequency being shaped inside another resonant frequency, both modulated independantly; so in that case, it really does become somewhat fractal. It's similar in concept to the idea of serial sync (where the waveshape resulting from a sync operation is used as the basis for another sync operation).

[lectrixboogaloo]

Gentleclock, I think it works something like this. The cz is only applying the shape (at key pitch) of one waveform to another higher pitched sine waveform. The pitch of the sine is linked to the value of the DCW and keytracking. Theoretically this should sound like the 'window' waveform with a strong harmonic. Casio's 'special' higher frequency 'multi-sine', of course, serves as the upper harmonic and window function(?) shape being the main oscillator sound. So there you go, your well emulated resonant peak, right? Sort of, under some circumstances. That's ok, though, because the CZ rez sounds are weird and not quite right in a good way.

The bazille version gives you further pitch control of the higher frequency 'multi-wave' (whatever it's called) and choose waveforms other than sine. With that, you can do strange bazillion things and also a very nice emulation of osc sync, and osc sync tends to make lots of vocal sounds if dialed in so.

But in a formant osc, does not one component remain pitch stable while the other tracks the keyboard? On the cz and bazille, I don't think that is so. Both components sound like they are tracking the keyboard. So, maybe there is some incidental overlap in sounds that both can produce, but I don't think they really are the same at all.

A sort of formant osc setup vs. 'fractal rez'. in bazille:

https://www.dropbox.com/s/se4hw8bc87uuw ... 20osc..h2p

I think the difference is clear but maybe I don't know enough about the formant osc you are comparing to.



Hasty patch could use some work, sounding ok over a couple octaves for now.

[Sendy]

But in a formant osc, does not one component remain pitch stable while the other tracks the keyboard?

Theoretically, a formant is pitch-independant, but the term has been generalized in popular synthesis to mean any shiftable resonant peak, whether keytracking or not.

The fractal resonance is the same as what Tone2 are doing with their "Impulse Modelling" in Rayblaster (yeah really, modelling.... I know!), though since Rayblaster uses this technique as the entire synthesis vehicle, they expanded it with some very cool extra parameters (formant and pitch noise, and the choice to have each impulse cut the other off, like in hardsync, or to let them overlap, as in granular synthesis, for example, giving an interesting middle ground between sync and granular technologies).

I prefer to call it windowed sync or smoothsync, since the windowing removes the gnarly irregularities you get with normal sync (the tearing you hear each time a new wavefront appears). You can set this up with any good synth with both sync and ringmod (or even filter mod will work).

To prove CZ resonance is just a synced sine with amplitude shaping, it's possible to "hack" the CZ synths with SYSEX data and turn off the windowing function of the resonant waveform. In that case you just hear a simple synced sine sweep, like you'd expect from any normal synth with a sine oscillator and sync.

Figure 19 on this page: http://en.wikipedia.org/wiki/Phase_distortion_synthesis and it's caption notes explain the process visually fairly well. Wave (d) is the windowing function, which in this case is derived from the phase accumulator by inverting it.

[Howard]

I prefer to call it windowed sync or smoothsync, since the windowing removes the gnarly irregularities you get with normal sync (the tearing you hear each time a new wavefront appears). You can set this up with any good synth with both sync and ringmod (or even filter mod will work).

Should be possible in ACE, then. IIRC Urs already explained to me how, but either I had a bad day or forgot since then. Do you have ACE, Sendy?

[Sendy]

Yes, it works with ACE, though you have to play around with the waveform mix to get the null point to line up with the discontinuities in the sync slave, since ACE's oscillators don't have phase adjustment. (Well, the LFO's do, but these can't sync eachother).

It's also possible to do smoothsync in DIVA now that the digital oscillator has been added - it's the only oscillator model in DIVA to offer sync and ringmod together. Again, you have to play with the waveshape a bit to get the zero crossing to line up.

It's also possible in Zebra but you have to defeat Zebra's DC blocking in the oscillators. I believe the patch is in one of my contributions to the patch bank.

[pdxindy]

Yes, it works with ACE, though you have to play around with the waveform mix to get the null point to line up with the discontinuities in the sync slave, since ACE's oscillators don't have phase adjustment. (Well, the LFO's do, but these can't sync eachother).

It's also possible to do smoothsync in DIVA now that the digital oscillator has been added - it's the only oscillator model in DIVA to offer sync and ringmod together. Again, you have to play with the waveshape a bit to get the zero crossing to line up.

It's also possible in Zebra but you have to defeat Zebra's DC blocking in the oscillators. I believe the patch is in one of my contributions to the patch bank.

I appreciate your level of knowledge... I frequently learn something from your posts!

[lectrixboogaloo]

I prefer to call it windowed sync or smoothsync,

Well, that sounds much better than 'multi-wave' or 'Casio's special multi-sine'.

Never really thought of the upper sine as being intentionally 'synced' as such, just some unfortunate side effect of a cheap way of stuffing an extra oscillator sound in to additively create a movable partial with. But, yes, I guess there really is no difference, is there?

since the windowing removes the gnarly irregularities you get with normal sync (the tearing you hear each time a new wavefront appears).

So, in hacking around the sync 'bug', a revolutionary new type of synthesis was born?

[xh3rv]

So, in hacking around the sync 'bug', a revolutionary new type of synthesis was born?

I get the impression filter resonance was how it was marketed, at least - it's kind of a backwards way getting there, windowing sines rather than rippling e.g. a saw into self-oscillation with a filter. Surely another critical motivation was a bit of inspiration dealing with the limits of digital technology at the time. It's an interesting if technical story, IMHO ...

A 'phasor' is simply a digital ramp wave, but is sort of the fundamental form generated in real-time by CZ/DX-style synths. It's very easy, and cheap in terms of digital circuits, to execute. Each successive sample is the previous value plus a constant value that depends on desired frequency - slower or faster frequency just means a smaller or larger increase per clock cycle. Exceeding maximum value simply wraps around back to minimum. (This is sort of how add/multiply digital arithmetic naturally plays out, and is often a computational error, but perfectly useful here.)

DX-7 users have a clear objection here - they only ever got sines! The next step: these simple ramps are transformed into sine by use of a fast memory lookup table with a sinusoidal waveform; in other words, the phasor indexes a sine. The table, perhaps several thousand entries, has a layout such that more or less one can look up a position given by the 'phasor' and get a sine. A single cycle of 'phasor' results in a single cycle of a sinusoid (i.e. phasor min = 0, phasor max = 2pi). With a phasor of a given frequency and a lookup table, this most of the way towards CZ/DX-style oscillators (interpolation, and a few range tricks are notable).

Phase distortion is simply passing the phasor, rather than straight to the lookup table, first through something that distorts it. For CZ stuff, Instead of a clean ramp, one ends up with a few ziggy angles. Ultimately this process equates to a fairly small number of lines of code, doing some clear, simple arithmetic. Minimal components are required to implement this.

Resonant waveforms are a result of a couple tricks. To get, say, 4 sines out of one phasor, just multiply by 4 (and wrap around) right before the lookup table. The window functions are very clever. Note they're all digitally clean - if one has the sines after the lookup table, one simply needs to do a few arithmetical operations on the original phasor to get the saw, triangle, or half-max/half-saw windows! Not much more than multiplying the multi-sine lookup table output and the post-transformation results in the final 'res' output.

At the time, each digital component in a setup like this was a precious resource. The same is true today but thanks to Moore's law it's exponentially cheaper, in a certain sense. Recycling the phasor (perhaps with the circuits already available for phase distortion?) is astonishingly economical, as well as sonically useful.

[e] - For fixed formants:

With fractal resonance, modulate with keyfollow at -5.00. This is pretty close to fixed frequency.

With Res distortions ... not so sure . Best I've figured out is to use an MMap. Fill the map with the 'quadric' function, set it's index to keyfollow (Key + 128 steps, or map/keyfollow IIRC both work) then modulate PDModDepth. I think -50.00 or so is a starting point but, it depends and it's not locked down across octaves very well. Probably there's a better way, not sure what the formula for how PD depth -> #cycles/ harmonics is.

[lectrixboogaloo]

So, in hacking around the sync 'bug', a revolutionary new type of synthesis was born?

I get the impression filter resonance was how it was marketed, at least

It was. From what I remember that's about all the cz manual says about it. I was really referencing what Sendy said about Rayblaster though. It is all more interesting than I thought it would be.

Phase distortion is simply passing the phasor, rather than straight to the lookup table, first through something that distorts it.

So, easy enough to play around listening in real time and visually. Read map gen 1 with a saw (uhm, ramp) lfo then plug map gen 2 into audio and put a sine in it and read it with map gen 1. Manipulate 'distortions' by changing shapes in map gen 1. Basically, same idea?

[xh3rv]

So, in hacking around the sync 'bug', a revolutionary new type of synthesis was born?

I get the impression filter resonance was how it was marketed, at least

It was. From what I remember that's about all the cz manual says about it. I was really referencing what Sendy said about Rayblaster though. It is all more interesting than I thought it would be.

Yeah, Sendy is entirely correct about this stuff - the overlap with some other synthetic methods is unmistakable. There's also VOSIM and FS1R type stuff on some of these tangents (some patents still around, even ... wondering how many patent violations Bazille is capable of )

Phase distortion is simply passing the phasor, rather than straight to the lookup table, first through something that distorts it.

So, easy enough to play around listening in real time and visually. Read map gen 1 with a saw (uhm, ramp) lfo then plug map gen 2 into audio and put a sine in it and read it with map gen 1. Manipulate 'distortions' by changing shapes in map gen 1. Basically, same idea?

Right layout, I think - I didn't mention there's a neat trick where the sine table is actually only 1/4th of a sine, since this can be flipped around (horizontally, vertically, or both) and tiled. This is less sampling error for the same amount of memory. Not sure, but I think the CZ plays with half a cosine (and some more trickery with the 2 slots for waveforms, gluing together the distortion).

To take the red pill and see the phase distorting, use a tap map with a ramp wave. Then it's a phasor (ramp wave) indexing a ramp wave table, and the intermediate distortion step will show up on the oscilloscope

[Urs]

wondering how many patent violations Bazille is capable of )

Pretty much anything one can do in Bazille one could do with 30+ year old hardware. Therefore any claim of patent infringement arriving at our door will be met with an example of prior art. Hehehe, I can't wait to crowdfund a movement for a patent dismissal.