Simple questions on synth tech basics...

[antto]

then how come you expect the filter to not be used when env_amount is zero?

anyway..
doing a square or pulse out of 2 sawtooths (one inverted) is an old trick.. it's perfectly valid
for example, if you're using mipmaps (or pretty much any other bandlimiting scheme) for a sawtooth - it might be easier to reuse it for the square/pulse instead of doing dedicated mipmaps (or whatever) for the square
you just process a sawtooth twice per sample to get a square..
i'm not familiar with Sylenth, can't comment on it

the gibbs thing is sign of steep resampling filter (or steep bandlimiting in the case of mipmaps)
the brightness of a square wave doesn't depend on whether it has gibbs ripples or not
but if the bandlimiting/resampling filter isn't very steep or it intentionally avoids producing gibbs ripples - then the resultant sound will not be as bright as normal
however, a very bright square wave at the "right" frequencies wouldn't produce ripples either, and will be as bright as possible
in other words, say we use the steepest possible brickwall filter (sinc interpolation) to generate a square
it will produce ripples most of the times, but it won't produce any ripples at frequencies where the period length is a whole integer number of samples, and the phase happens to reset at 0.0 exactly.. then the square will look as sharp as a naive square

[fluffy_little_something]

I don't expect the filter not to work, I know now that it still works. But I think it is not intuitive when nobody explains it to you, unlike with the amp envelope, whose function is pretty obvious. So for quite some time I did not understand what was going on. I just used those filter controls based on the sound effect they have based on experimentation, not in a constructive way.

Took a screenshot of two pulse waves, the one on the right sounds better to my ears. All my other synths' pulses look like the one on the left, though.

https://app.box.com/s/lzjjrrl42zd3d8k3xw07


"you just process a sawtooth twice per sample to get a square."

Can that cause pw modulation to fail? I tried it and it doesn't work well, there are drop outs and other irregularities.

[camsr]

If each sawtooth is in phase with another, they will cancel!
One sawtooth out by 50% phase gives the square.
Use frequency modulation on one of the oscillators to simulate PWM.
You will have to limit how much FM occurs if you want to prevent the in-phase interference (sounds backwards )
If the oscillators are the same wavetable, and they are linear ramps, you will get a square PWM.

[fluffy_little_something]

Strange, yesterday I tried pwm by modulating the phase, and it sounded crappy. Now I tried the very same thing and it sounded perfect

[fluffy_little_something]

What does "bandlimited" osc mean and what is it good for?

[antto]

What does "bandlimited" osc mean and what is it good for?

forget about digital and analog for a moment, and think about a sawtooth wave..
a mathematically perfect sawtooth, with an infinitely sharp edge.. that would have an infinite number of harmonics (that's how you get the sharpness of the edge)
now, an analog sawtooth can't have an infinitely sharp edge, but it can still be pretty pretty sharp.. way sharper (with more high harmonics) than what you can capture in a 44.1kHz digital signal

if you generate a sawtooth digitally via adding together the individual harmonics - if you keep adding more and more harmonics - you will get closer and closer to a perfect sawtooth (the edge will be getting sharper and sharper) but you have to stop if you reach (or exceed) nyquist, otherwise you will break the rules of dsp and get aliasing
well, that's it.. you have to generate a bandlimited sawtooth, limited between 0Hz and Nyquist.. your sawtooth shouldn't attempt to have content above nyquist because that's not possible and any such content gets mirrored into the [0 - nyquist] band in the form of aliasing

and when the analog sawtooth (which has a pretty pretty sharp edge) has to be sampled digitally - the ADC must also apply a steep lowpass filter before actually sampling the signal..

[fluffy_little_something]

Thanks, I even understood So, when sampling an analog saw the lp filter is necessary because the ADC does not have a sufficiently high sampling rate that would be twice as high as the highest frequencies in the analog signal, which I suppose could be way beyond audible frequencies?

I used to have a synth (one of those big Chris Kerry synths) where one could specify the number of harmonics, the max was 512 if I remember correctly. Is that a lot? Does it allow for those "sharp edges"?

[antto]

Thanks, I even understood So, when sampling an analog saw the lp filter is necessary because the ADC does not have a sufficiently high sampling rate that would be twice as high as the highest frequencies in the analog signal, which I suppose could be way beyond audible frequencies?

it doesn't matter if the ADC had a sufficiently high sampling rate or not, nor wether it "knew" what's the highest frequency that the signal will contain
at the end of the day, you want to record your sound at a sampling rate of 44.1k or 48k, or 96k, etc.. and the recorded signal should not contain any content above nyquist
and the only way to do this is to filter out the content above nyquist

I used to have a synth (one of those big Chris Kerry synths) where one could specify the number of harmonics, the max was 512 if I remember correctly. Is that a lot? Does it allow for those "sharp edges"?

well, 512 harmonics is exactly 512 harmonics.. whether it's enough or not depends on a few things
- if you want to do a sine wave - yes, it's enough.. even overkill
- if you want to do a periodic waveform that hass not more than 512 harmonics - then it's perfectly enough
- if you want to do one of the basic shapes in subtractive synthesis (saw, square, etc) then it might not be enough when that waveform is going to be played at a very low frequency
play a sawtooth thru a spectroscope and check how many harmonics it has from the fundamental all the way to nyquist... notice that as you play a lower pitched note the fundamental gets closer to 0Hz and the distance to nyquist becomes greater, and the spacing of the harmonics gets smaller
so, in that case, to maintain the sharpness of the edge(s) of your sawtooth, the lower the frequency it oscillates at - the more harmonics you will have to have till you reach nyquist
with a limit of 512 harmonics max - there will be a given frequency below which your sawtooth will start losing its sharpness

[fluffy_little_something]

That is one reason why I like Sylenth, the saw sounds sharp even at very low frequencies. I just checked what the lowest note is that it can produce, it is C1 (the peaks are so few one can count them by ear ) and the peaks still sound bright On my SE synths C1 sounds more like muffled knocking, almost like a bass drum even.

[Aaron Zilch]

This seems like a fun little discussion.

Fluffy, if you are enjoying Sylenth you should check out Hive, it's in open beta so it'll be free for you to experiment with till they get around to the full release ( which I'm personally hoping they push back till after their vacation ). It's mipmapped, but mipmapped by U-he. So they freely admit to cutting the corners they are cutting, because they normally don't cut them. And it still sounds brilliant..becuause….U-he



Diva is their analog modeled flagship and it cuts no corners and sounds amazing. I'm pretty sure the free TyrellN6 is using bits of the same engine ( but I'm sure someone here will correct me if I'm wrong )

Of course as a result, they chew up CPU. And kids are like " NOOOOO!!!! dis sux!!! I can has 30 wubs in Massive on mi sweet Dell. Why U-he suck at thee programming?!?!? It sound no better on my headphones and they are PRO studio by dre. BY DRE!!!"

So now they are putting out Hive, and since Lenner have dropped the ball on 64bit, the EDM punters are going to line up and throw money at someone who truly deserves it, for a synth where they didn't even have to break a sweat. It's hilarious. "Oh you don't appreciate true digital synthesis? Well here's a much better version of what you consider passable. Your welcome."

So that covers the dirty secrets of DSP oscillators…next up you should start comparing all the bullshit ringy digital filters out there to Cytomic's The Drop.


…and then you can join me on the streets trying to find homeless that look healthy enough to get us some organ money for more processing power

[camsr]

Not all sawtooth oscillators are linear. Some will look curved, mimicking a capicitor discharge curve.

[peteulge]

And what's the point of this post except various u-he advertisement bullshit?!?

This seems like a fun little discussion.

Fluffy, if you are enjoying Sylenth you should check out Hive, it's in open beta so it'll be free for you to experiment with till they get around to the full release ( which I'm personally hoping they push back till after their vacation ). It's mipmapped, but mipmapped by U-he. So they freely admit to cutting the corners they are cutting, because they normally don't cut them. And it still sounds brilliant..becuause….U-he



Diva is their analog modeled flagship and it cuts no corners and sounds amazing. I'm pretty sure the free TyrellN6 is using bits of the same engine ( but I'm sure someone here will correct me if I'm wrong )

Of course as a result, they chew up CPU. And kids are like " NOOOOO!!!! dis sux!!! I can has 30 wubs in Massive on mi sweet Dell. Why U-he suck at thee programming?!?!? It sound no better on my headphones and they are PRO studio by dre. BY DRE!!!"

So now they are putting out Hive, and since Lenner have dropped the ball on 64bit, the EDM punters are going to line up and throw money at someone who truly deserves it, for a synth where they didn't even have to break a sweat. It's hilarious. "Oh you don't appreciate true digital synthesis? Well here's a much better version of what you consider passable. Your welcome."

So that covers the dirty secrets of DSP oscillators…next up you should start comparing all the bullshit ringy digital filters out there to Cytomic's The Drop.


…and then you can join me on the streets trying to find homeless that look healthy enough to get us some organ money for more processing power

[fluffy_little_something]

What exactly is your point there, Aaron? ^^

I tried both Hive and Sylenth simultaneously, and ended up buying the latter for several reasons.

There is a 64-bit version of Sylenth by the way...

[fluffy_little_something]

As I was playing around with portamento, I was wondering if it would not make more sense to have the portamento time depend on the distance between the two consecutive keys played instead of using a fixed time for all distances, be it two notes apart or 4 octaves.

[aciddose]

So you want linear vs. log glide.

Xhip has this if your other plugins do not.