Repro-1 (out now)

[e-modic]

I know this is all about finding the best Algo but beside this, thanx for great freeware. On my latest i7 I don´t see CPU hugs I just enjoy a damn good Synth. My preferred 1 is going to lose, a shame, maybe I am too much old school.

[Cyoon]

ok, better to recheck the filters at higher sample rates then.

[Urs]

ok, better to recheck the filters at higher sample rates then.

Well, they kind of always work at the same rate of 352/384 kHz...

[david.beholder]

grit and noise does not necessarily mean aliasing...

otherwise all distortion pedals are aliasing.

[Vesius]

Seems that some people confuse "aliasing" with distortions created when the mixer is overloaded and the result fed into a high resonant filter!

model 1 aliases with nothing going through the mixer. On a spectrum analyzer you can see 4 do nearly the same at very very high freqs, but honestly, that's outside my hearing, so I'm completely unbothered. So unbothered that I voted for it, I like the complex behaviour.

Not to be a PITA about this point, but ... it just does. I like the filter, but I don't like the aliasing.

[kybernaut]

Seems that some people confuse "aliasing" with distortions created when the mixer is overloaded and the result fed into a high resonant filter!

model 1 aliases with nothing going through the mixer.

I agree about Filter 1

[david.beholder]

model 1 aliases with nothing going through the mixer.
Not to be a PITA about this point, but ... it just does. I like the filter, but I don't like the aliasing.

Man, it's not alias and analog filters behave like this quite oftenl with high gain/high resonance.

[Vesius]

*double post* apologies.

[Vesius]

model 1 aliases with nothing going through the mixer.
Not to be a PITA about this point, but ... it just does. I like the filter, but I don't like the aliasing.

Man, it's not alias and analog filters behave like this quite oftenly.

I'm sorry, in this instance I think you're mistaken - the resonance of the (not perfect, quite correctly, ie it has overtones) sine hits the nyquist and starts coming back down into the audible range as the cutoff knob moves up. It's really classic aliasing.

If you can't hear it, then I'd suggest popping open a copy of SPAN, setting the window to 16384 and having a look - then take a look at an article or two on aliasing like this http://storm.uni-mb.si/CoLoS/applets/aliasing/ - the last image is the one to bear in mind here - as the cutoff (the loudest spike) moves up, the overtones hit the nyquist and literally bounce back and down into the audible range. This should give you a clearer idea of what to be listening out for.

Given that Urs has stated that his Pro1 goes to at least 68Khz, it is difficult to see what phenomenon might cause an analog synth to produce an entirely digital artefact, although, true, it is not enitrely possible to discount the possibility. Either way, I find something that is functionally equivalent to aliasing to be necessarily as undesirable as aliasing.

[david.beholder]

I'm sorry, in this instance I think you're mistaken - the resonance of the (not perfect, quite correctly, ie it has overtones) sine hits the nyquist and starts coming back down into the audible range as the cutoff knob moves up. It's really classic aliasing.

If you can't hear it, then I'd suggest popping open a copy of SPAN, setting the window to 16384 and having a look

Thanks, I don't need to be explained right now as I know what alias, underlying math and what Urs said over last weeks.
Please attach your screenshots and audio samples instead so we all could eat our hats or show you that you're wrong

[Cyoon]

isn't it the filter self oscillation bleeding through? Somebody mentioned that i think.

[Vesius]

Man, it's not alias and analog filters behave like this quite oftenly.

I'm sorry, in this instance I think you're mistaken - the resonance of the (not perfect, quite correctly, ie it has overtones) sine hits the nyquist and starts coming back down into the audible range as the cutoff knob moves up. It's really classic aliasing.

If you can't hear it, then I'd suggest popping open a copy of SPAN, setting the window to 16384 and having a look

Thanks, I don't need to be explained right now as I know what alias, underlying math and what Urs said over last weeks.
Please attach your screenshots and audio samples instead so we all could eat our hats or show you that you're wrong

No problem, filter goes between 4 and 7.5 roughly.

Synth settings:

synthsettings.jpg

Aliasing shown in Span:

spanalias.jpg

And a link to the soundfile - it's downloadable as a .wav, so you can hear it properly and not go through the soudcloud compression.

https://soundcloud.com/speakafreaka/filter1aliasing

Anything else?

[magickalmutagen]

It's between filter model #1 and #4 for me. I voted for #1 as I found it to sound both squelchier and crunchier. #1 is less muted than #4 and contained a little distortion (possibly aliasing, though I think it added to the sound). Out of #2, #3 and #5, it was #3 that came out on top.

[ATN69]

Number #5 has my vote. It cut the high end just slightly (not as much as #2) giving it a slightly warmer feeling. A good tradeoff between #2 and #1.

[kybernaut]

While Fiter 1 creates audible aliasing (as Vesius demonstrated quite nicely) the other models defeat these audible artifacts more or less effectively....

The art is presumable to preserve as much of the high frequency content as possible without letting digital artifacts through...

Probably Filter 4 is doing this best after all.... Seems that I was tricked as well by the high-frequency harshness and therefore voted for Model 3