Warmy's SuperSaw 1.0 beta

Warmonger · 12 Jul 2012

My first synth is ready. All known issues were fixed, GUI is reworked, sound is amazing Check this out and create some more presets 😉

DLL version: https://www.dropbox.com/s/dp97felnfqf9o ... perSaw.dll

Xtinct · 12 Jul 2012

Hi you need to turn the hold off in the filter ADSR as some sound can be heard after the amp ADSR has finished if the release on the filter is longer than the amp even if the filter ADSR amount is 0. odd 😕 Edit Most prominent when the envelopes shapes are at there lowest settings

cyto · 12 Jul 2012

After a cursory listen, it sounds pretty good!

But...

This thing is very prone to subnormals. My little netbook choked up almost immediately after releasing the first note. After a simple once-over, I didn't see any subnormal prevention measures in any of the filters. I feel partly to blame, because I didn't warn you of this with the HP thing for the reverb. But even turning the reverb off, I'm still noticing the CPU spikes. Do a search for "denormals" or "subnormals" and you can get an idea of different ways to prevent them. If I have time later, I may be able to go in and help you track them down.

-cyto

EDIT: See my latest post for a long-winded discussion about subnormals.

Warmonger · 13 Jul 2012

Thanks for help. I really forgot about denormals, as this reverb was reworked over and over, leaving nothing from the original 😉 Also, now I see why synth interrupted terribly when presented on laptop :P As to envelope... well, it may explain weird behaviour of sound under certain setings. I thought these effects come from feedback loops, but apparently it was something very diferent.

Warmonger · 16 Jul 2012

Hi you need to turn the hold off in the filter ADSR as some sound can be heard after the amp ADSR has finished if the release on the filter is longer than the amp even if the filter ADSR amount is 0.

I did that and didn't really notice a difference. However, now I see that strange sudden echo occurs for sounds with non-zero pre-delay time on reverb. It may be some tricky technical issue, or just the fact that reverb algorithm can't work properly without other sound flowing 🙁

EDIT: Attaching schematic after fixes.

infuzion · 21 Jul 2012

Seems neat, but a few comments: Your deflanger does nothing for randomizing individual notes for a chord. Is this by design? Using Code for OSCs will cause alising. May be good at times, but people tend to avoid that. SuperSaws have been built with SM's OSCs, which help reduce this issue.

Warmonger · 21 Jul 2012

In the other thread people suggested that stock OSC is band-limited, so I swapped to code to avoid that. In fact, the whole idea of Supersaw is to handle aliasing in a good way.

Your deflanger does nothing for randomizing individual notes for a chord. Is this by design?

I don't have other idea to simulate free-running oscillators. Any advice would be welcome.

infuzion · 22 Jul 2012

Warmonger wrote: In the other thread people suggested that stock OSC is band-limited, so I swapped to code to avoid that. In fact, the whole idea of Supersaw is to handle aliasing in a good way.

Your deflanger does nothing for randomizing individual notes for a chord. Is this by design?

I don't have other idea to simulate free-running oscillators. Any advice would be welcome.

Band-limiting is usually considered better; search for Nyquist.

I would use a Float Array to Poly primitive, & randomize that array about 1 second on loading the VST.

Warmonger · 22 Jul 2012

Band-limiting is usually considered better; search for Nyquist.

Care to explain? Band-limiting, not Nyquist of course.

All I can tell is that after replacing default oscillators, sound got better.

randomize that array about 1 second on loading the VST

The issue is to not reset oscillator phase on every key press. So I change initial phase of one oscillator before it gets reset to that value.

infuzion · 22 Jul 2012

Warmonger wrote: Band-limiting is usually considered better; search for Nyquist.

Care to explain? Band-limiting, not Nyquist of course. All I can tell is that after replacing default oscillators, sound got better.

https://www.google.com/search?q=Band-limiting+Nyquist & search the forum Band-limiting does "dull" the sound a bit, that is because you have reduced the high freq. You also have to realize that in (pre)mastering, there will be filtering out those high freq, so you are eating up space & energy in the mix, which will make your synth not play well with other sound sources.

Warmonger wrote: randomize that array about 1 second on loading the VST

The issue is to not reset oscillator phase on every key press. So I change initial phase of one oscillator before it gets reset to that value.

hmmm, I wonder if it is even needed then? Or may cause the issue (nulling) you are trying to prevent (phasing).

martinvicanek · 22 Jul 2012

Warmonger wrote: Care to explain? Band-limiting

Band limiting is introduced in harmonically rich waveforms to suppress frequencies above Nyquist and hence avoid aliasing. To demonstrate the effect of band limiting, I set up this little schematic. You can hear the aliasing of the (non-bandlimited) saw when modulated at high pitch, there are spurious frequencies moving in the opposite direction of the fundamental. That's because in the sampling process, frequencies above Nyquist are mirrored back into the audible range. There is no aliasing for the band-limited saw.

Warmonger · 22 Jul 2012

Ok, I know that. Still, in the synth there are two (different) ways to prevent aliasing, namely key-following high-pass filter (as in orignal supersaw synth) and PolyBLEP function which reduces aliasing at phase-shifting, especially layered saws. Neither of these seems to be a problem here.

I think you should focus more on actual rather than virtual problems :P First rule of engineering: don't fix that which was not broken.

Also, here's first version of Manual, it may enlighten you: https://www.dropbox.com/s/i59l4x1rfk582t2/SuperSaw.pdf

hmmm, I wonder if it is even needed then? Or may cause the issue (nulling) you are trying to prevent (phasing).

Well, certainly at some point oscillators may stack so that they sound worse than expected. The question is, how often it happens and how anoying it becomes. Getting one phased note is not that bad as getting every note phased in same way.

Still, random phase makes synth sound more natural and a bit different every time.

infuzion · 23 Jul 2012

Warmonger wrote: Ok, I know that. Still, in the synth there are two (different) ways to prevent aliasing, namely key-following high-pass filter (as in orignal supersaw synth) and PolyBLEP function which reduces aliasing at phase-shifting, especially layered saws. Neither of these seems to be a problem here.

Ahha, I did not see those modules! I assumed anti-aliasing would be inside the OSC modules, not outiside; my mistake. Hmm, I do see your readouts, but some how my imagination is seeing/hearing a bit of aliasing. I might be paranoid from when I last tried to make new OSCs: [old link]

Still, random phase makes synth sound more natural and a bit different every time.

I agree!

Thanks for your time & OSM. Honestly I think this has commercial possibilities. $

Warmonger · 2 Aug 2012

Supersaw is live on KVR!

http://www.kvraudio.com/product/warmys- ... -warmonger

Technical issues have been fixed, knob readouts were improved, some generic presets added. What is there not to like?