KVR Audio

Taifunk · Post by **Taifunk** » Thu Apr 23, 2015 3:02 pm

really nice

forgive my curiosity you can post an audio example?

Taifunk · Post by **Taifunk** » Thu Apr 23, 2015 3:14 pm

double post sorry

Mutant · Post by **Mutant** » Thu Apr 23, 2015 4:14 pm

Taifunk wrote:really nice forgive my curiosity you can post an audio example?

https://www.sendspace.com/file/48ehqo
OFC the osc texture, filter sound and the effects will be different, but IMHO the curve is almost perfect.

Strings chords made with Xhip.
http://xhip.net/synth/

Taifunk · Post by **Taifunk** » Thu Apr 23, 2015 6:40 pm

thanks!

wow it sounds perfect

Xhip is one of my favorite synth I use it always.
if not mistaken on the real machine the aggressive filter is the CEM3320
I have something similar in the AJ OB-Xtreme I'll try to recreate the sound with this.thanks again.

aciddose · Post by **aciddose** » Thu Apr 23, 2015 7:04 pm

I have no idea how to interpret the graphs you've posted. Can you post these with the X/Y axis defined and a description of what is in the graph. Is it an attack followed by a release?

What is plotted? The complete spectrum? Only the oscillation of the filter?

Knowing the Z/color axis would also be nice. Are we seeing dB? Linear?

aciddose · Post by **aciddose** » Thu Apr 23, 2015 7:23 pm

As far as I can tell after staring at it for a few minutes is that the Y (vertical) axis is time. X (horizontal) is frequency and the plot appears to be in dB with black = -40 or so?

You want envelopes with an attack asymptote of 100% or very near to 100%.

Xhip has an asymptote of 200%. This can be adjusted although there is currently no parameter for it.

This means the shape of the attack is actually the first 1/2 of the log curve you'd normally get. An attack must have an asymptote greater than 100%, otherwise it will take an infinite amount of time to finish. (Linear = infinite asymptote.)

Most software has a fully linear attack, which in my opinion sucks. This is due to the fact the authors do not understand what an asymptote is, and they had heard at some point that synthesizers such as the SH-101 use a "linear" attack without understanding that it is in fact a 200% asymptote.

I'll upload an alpha version of Xhip with the asymptote set to 102%. This should be close to what you have in the graphs.

http://xhip.net/temp/xhip_alpha_102_asymptote.7z

Just remember:

This is 100% alpha, expect things to go wrong. (Although it should work okay.)
The presets won't be compatible, don't save presets with this version and expect them to sound the same in another! They should load, but the attack asymptote will be different.
I think there are some problems with buttons getting stuck in the GUI code, just click left/right on the GUI background to reset the button state if this happens.
Comes with a new default "factory" bank, right click on the buttons for other options. Right click on the text entries to edit. For example, right click save-bank for an option "save as default" so it will automatically load in new instances.

Mutant · Post by **Mutant** » Thu Apr 23, 2015 8:29 pm

aciddose wrote:I have no idea how to interpret the graphs you've posted. Can you post these with the X/Y axis defined and a description of what is in the graph. Is it an attack followed by a release?

What is plotted? The complete spectrum? Only the oscillation of the filter?

Knowing the Z/color axis would also be nice. Are we seeing dB? Linear?

Yes i believe it plots complete spectrum and log scale.
Y = time, X = frequency, color = dB
It is the freeware Dust Analyzer set to sonogram mode and "stereo power" something mode.
http://www.kvraudio.com/forum/viewtopic.php?t=380389
(and i had to google "asymptote")

aciddose wrote:You want envelopes with an attack asymptote of 100% or very near to 100%.

Xhip has an asymptote of 200%. This can be adjusted although there is currently no parameter for it.

This means the shape of the attack is actually the first 1/2 of the log curve you'd normally get. An attack must have an asymptote greater than 100%, otherwise it will take an infinite amount of time to finish. (Linear = infinite asymptote.)

Most software has a fully linear attack, which in my opinion sucks. This is due to the fact the authors do not understand what an asymptote is, and they had heard at some point that synthesizers such as the SH-101 use a "linear" attack without understanding that it is in fact a 200% asymptote.

I'll upload an alpha version of Xhip with the asymptote set to 102%. This should be close to what you have in the graphs.

http://xhip.net/temp/xhip_alpha_102_asymptote.7z

Wow ! Thank you very much

Ill install and check it ASAP.

aciddose · Post by **aciddose** » Thu Apr 23, 2015 8:54 pm

I'll add it to my list of features to add in the next version. I just want to get the current preset "stable" to ensure it is compatible with the last version (7) and then I can start adding or modifying all sorts of things.

The "const attack" switch is a similar feature which should only rarely be used. I've been thinking maybe to add an "advanced" page flip button/tab/corner to the envelope and modulator sections for these sort of parameters. I've always felt the asymptote would be only very rarely used, and so it was hard to justify putting it on the GUI. One other major issue is the fact that Xhip is currently set up to provide all parameters as automation targets, which might lead to the list being a bit too long with these rarely used parameters.

Still though, I definitely appreciate that the parameter will be used or needed once in a while and it certainly does provide an important effect, otherwise it wouldn't matter what % asymptote were used at all.

The proper name is "horizontal asymptote". Now you know what to call it

I just use "asymptote" as the "horizontal" should be inferred from the context.

Mutant · Post by **Mutant** » Thu Apr 23, 2015 8:55 pm

Almost there

aciddose · Post by **aciddose** » Thu Apr 23, 2015 8:59 pm

That isn't the alpha I linked, that's the 200% asymptote from the standard version. Make sure you use "Xhip alpha" and it says "Xhip alpha 8.1.582 102% attack asymptote!" on the screen when it starts up.

Mutant · Post by **Mutant** » Thu Apr 23, 2015 9:03 pm

Yes it is the alpha. I have it open right now.

aciddose · Post by **aciddose** » Thu Apr 23, 2015 9:15 pm

Hmm, well I suppose the graph may be sharper than I thought. Is it in decades? I suppose it is, as the harmonics should be spaced evenly and they're all crunched together on the top end.

So you need an asymptote which is extremely flat, actually so flat as to be nearly useless for any other purpose.

I see the OB-Xa uses the CEM-3310 for the envelope.

Now, the problem is the manual for the CEM-3310 clearly specifies the attack asymptote:

Peak:
min: 4.7
typ: 5.0
max: 5.3

Asymptote:
min: 6.1
typ: 6.5
max: 6.9

So, in percentages (asymptote / peak):

min/max: 115%
typ: 130%
max/min: 146%

(Note the datasheet specifies max tracking error for asymptote = 4%, so using min/max and max/min is way out. The values are all within 4% of 130%, so 126% to 134%.)

So there is no way that graph is simply the result of the envelope. Instead, it seems that van halen's OB-Xa was severely out of tune. I'd attribute this to a horrible tuning of the filter scale trimmer.

So, we want to start with asymptote 130% and then adjust the filter tracking. Now the problem is the keytracking will no longer work, it will be equally out of scale.

Mutant · Post by **Mutant** » Thu Apr 23, 2015 9:30 pm

aciddose wrote:Hmm, well I suppose the graph may be sharper than I thought. Is it in decades? I suppose it is, as the harmonics should be spaced evenly and they're all crunched together on the top end.

I don't know how the Dust Analyzer works, i just split the original and VSTI channels left and right and compare the green and red curves.
And i can match the curves and keytrack response best with Elektrostudio Model Mini in exponential mode.
http://www.elektrostudio.ovh.org/?go=mini

aciddose wrote:So you need an asymptote which is extremely flat, actually so flat as to be nearly useless for any other purpose.

I see the OB-Xa uses the CEM-3310 for the envelope.

Now, the problem is the manual for the CEM-3310 clearly specifies the attack asymptote:
Peak:

min: 4.7

typ: 5.0

max: 5.3

Asymptote:

min: 6.1

typ: 6.5

max: 6.9
So, in percentages (asymptote / peak):

min/max: 115%

typ: 130%

max/min: 146%
(Note the datasheet specifies max tracking error for asymptote = 4%, so using min/max and max/min is way out. The values are all within 4% of 130%, so 126% to 134%.)

So there is no way that graph is simply the result of the envelope. Instead, it seems that van halen's OB-Xa was severely out of tune. I'd attribute this to a horrible tuning of the filter scale trimmer.

So, we want to start with asymptote 130% and then adjust the filter tracking. Now the problem is the keytracking will no longer work, it will be equally out of scale.

Ill download that youtube tutorial video i linked earier and check the curve there, if 2 synths have the same curve then the spec sheet must be wrong.

Mutant · Post by **Mutant** » Thu Apr 23, 2015 9:42 pm

Mutant wrote:Ill download that youtube tutorial video i linked earier and check the curve there, if 2 synths have the same curve then the spec sheet must be wrong.

No go... looks like YT cuts everything at about 15.7Khz so the very top of the resonance sweep is not visible.
But before reaching the top, the curve that begins at 1:44 looks just as it looks in Van Halen - 1984.

[edit] But the alpha version you uploaded today is good enough.

Would be nice to have a slider controlling the curve, but i can wait for that.

[edit2]See ? Green is Xhip, Red is Van Halen.

Not 100% there, but close enough for happiness.

Thank you very much again.

fluffy_little_something · Thu Apr 23, 2015 9:56 pm

I also had a problem understanding the diagrams. It is unusual to have time on the y axis. The x axis is traditionally the time axis in many applications.

Give me NONlinear ADSR !