Osc based on Martin's WaveShape

[k brown]

Just thought I'd post this osc built around a pair of the WaveShape oscillators.

Many of you have probably sussed out much of what's in here yourself already, but here it is in 'convenient form'.

- A pair of the WS osc configured to output Saw and Pulse simultaneously.
- A sine is mixed into the output via the Timbre control, so as the Timbre is increased, the sine is mixed in at increasing level, to keep up the strength of the fundamental, which otherwise gets weak at high settings of the Timbre control.
- A bit of maths keeps the Pulse output roughly centered as the Pulse Width is narrowed. PWM is best done with a unipolar Triangle LFO.
- A selector inside can be used to switch the Pulse output between Pulse, and a pair of Saws that can be tuned to intervals using a second Detuner connected to the 'Freq2' input.

A useful addition to this is separate Timbre controls for the two WS oscs (even separate phase mod inputs - try 'FMing' only one of the oscs used to make to pulse wave!).

I just love the sweet sound of Martin's WS osc when the Timbre is wept by an envelope. It also takes FM (phase mod) very nicely.

Any refinements to this would be most welcome! (very likely I've done some of this a bit clumsily)

WaveShape +.fsm

(4.55 KiB) Downloaded 654 times

[R&R]

Feels like everything essential needed packed into one module... kinda almost with filtering as well

I don't understand the pulse centering which offsets into DC, but then again i'm me...

[k brown]

The PW knob used should range 0.5 to 0.9 (or 0.95); not 0-1.

[R&R]

Maybe it's just me not understanding the context of it's usage. This is how I hooked it up to test...

I'm not that versed in these matters so...
Guessing you offset to keep the waveform shape (shapes peaks) intact, or you might be thinking all along of using a DC blocker for this osc. So the sum (final output) has to be adjusted?

[R&R]

Still don't know the intended function of the DC offset

This is how i'd use this module probably though.
Just added simple mapping (or what some call it) on the timbre input.
Whether it's suitable to run modulation thru the mapping, or directly, is probably to own taste. If not modulating thru the mapping the "mapping poly's" can be translated into green for use with know only, to save CPU instead I guess.

Maybe there is a scientific way of mapping according to "average amp of timbre/spectral content" or something... of the Waveshape Osc but that's on MV level of math

[k brown]

First, a correction:

The green 'Timbre Knob' float input is to be connected to the float output of said knob, not a separate one (that's why it's labeled 'Timbre Knob').

The only reason there is a stream and a float connection for the Timbre knob, is that the module's Timbre input must be stream so it can be modulated (and therefore best connected to the Timbre knob through a DeZip). The mix-in of the sine works best if the control has a quasi-log taper, rather than linear, so that is connected to the Timbre knob's float out via a float multiply. I tried using the DeZipped Timbre knob connection through a stream multiply, but it didn't seem to work the same.

WaveShape_Timbre connection.fsm

(182.07 KiB) Downloaded 639 times

I see what you mean as far as DC offset; with any PW other than square, the wave drops below center when Timbre is reduced toward sine. I was just focused on keeping the pulse waves centered when Timbre is near full, and the PW knob moves from square to narrow. Without the 'centering maths' the opposite happens: sine remains centered, but as the PW is moved from square to narrow, the waveform shift up above the center line. Being an audio oscillator and not an LFO, I'm not sure it makes a difference. Is one worse than the other? Does either matter? I don't know, may be someone else will chime in (if there still IS anyone else). But you're right, with a DC blocker on the Pulse output, everything will stay centered.

I'm not understanding the advantage of the exponential 'mapping'.

[R&R]

The green 'Timbre Knob' float input is to be connected to the float output of said knob, not a separate one (that's why it's labeled 'Timbre Knob').

Got it! Now it makes sense

I tried using the DeZipped Timbre knob connection through a stream multiply, but it didn't seem to work the same.

Seems to work the same? What's different, or at what settings/usage case does it behave differently?

I'm not understanding the advantage of the exponential 'mapping'.

Primarily a matter of personal taste/design choice or newbie mistake/bad choice from my part perhaps
I try to avoid knobs that have most perceived effect at for example"1/4" of it's rotation, even if it makes sense
to math and scale. Makes for better distribution across rotation for micro tuning of the knobs.

But doing it this way can cause some headache For example as I have opted out of the regular tracking and also done some remapping of my filter cutoff:s/reso:s in my newbie synth. I haven't accounted for that "yet", but will probably look at it closer later on.
Feels it can make a synth quite unique with divergent behaviour of knob interaction vs modulation. It might annoy the hell out of synth purists maybe

...with a DC blocker on the Pulse output, everything will stay centered.

I seem to have alot of experimental BS in my newbie synth, being new at FS etc ...so I have been forced to use some extra DC blockers and slew limiters. But try to avoid it if I can preadjust.

may be someone else will chime in (if there still IS anyone else).

Low activity here and I'm not much of a knowledgeable exchange unfortunately...

[Tepeix]

For the exponential mapping, it depend of the effect, but sometime the low value contain a lot more difference,
for filter id like to simply use a x^2 (i did not try more for now).
Specially when there's an lfo connected the remapping could help, or the bass will be swept ultra rapidly against the high.
If no lfo, would be better to do this in green before the zip.
Yet i would prefer to use another formula, i don't know if it change a lot. I prefer to avoid variable if possible.
So i will do h=x*x; y=h+h-(h*h).

That's some cool and interesting sound we have here !!)

[R&R]

Specially when there's an lfo connected the remapping could help, or the bass will be swept ultra rapidly against the high.
If no lfo, would be better to do this in green before the zip.

I agree sometimes it can be beneficial with some remapping... and sometimes it's best kept in original input/linear.

Since i'm personally a moron at math, and not working to match to any scale or convention... I take all sorts of liberties even using Sin() for some greens. But when in poly I often combine x^2 with inversion to get S-shapes.
But all this depends on the approach I guess.
My newbie synth is very knob-interaction oriented so this takes precedence over modulation outcome.
I also try to avoid any deadzones on my knobs if possible But I have some exceptions where I intentionally lower/overshoot certain outputs, to create tiny deadzones.
Quite a mess one could say

[R&R]

Hmmm. I recognize that shape

Shape_and_a_wave_we_go.jpg (50 KiB) Viewed 15599 times

Not nearly as sweet sounding and also somewhat transitory in my synth though...

I am not using the waveshape osc at all... but instead one of his other osc's, so I guess I have a ghost in my newbie synth. MV is haunting my synth.
Spooky action at a distance I think Albert called it?

It might be the hundreds and hundreds of hoping dezippers of his