KVR Audio

Markku · Post by **Markku** » Thu Jul 13, 2017 7:13 pm

Xhip runs perfectly in Linux... with 32-bit Wine.

It is actually very usable due to being light. This system (32-bit Ubuntu) runs synths like Sylenth1, Zebra and even Diva and Bazille, but not many instances; with Xhip I don't need freezing tracks practically ever.

But I guess nobody cares. Linux is constantly ignored.

yul · Post by **yul** » Thu Jul 13, 2017 8:33 pm

I would love to use Linux but's it's just too challenging and time consuming to set up partitions, drivers etc no? Last time I checked I would probably need a whole week of research and trial/error for me to get going.

Back on topic, Xhip is just fabulous!!

aciddose · Post by **aciddose** » Thu Jul 13, 2017 9:01 pm

Linux isn't ignored: but it is very difficult due to: "Wouldn't it be easier if the software did X?" -> "No, the software was hard to write so it isn't surprising that it is hard to use."

Not everyone is a linux geek who spends hours learning how to use the shell and memorizing different methods (tools and their parameters) to get things done.

That said; these days it is trivial to install a pre-configured version of one of the popular linux distributions which comes with a GUI like gdk/gtk/gnome set up. It is a little bit more effort to install and configure software but really not by much.

From a programmer's perspective: Linux is easily many times more difficult to use than Windows or OSX. OSX is many times more difficult than Windows. I'm referring to APIs here. For all the potential shortcomings of win32; it is an absolutely brilliant C interface that has stood the test of time and provided flexibility, ease of use and most if not all of the features used by millions of developers. This is for good reason though: Linux is designed as a modular system where the kernel doesn't get involved in things like window management. This is a lot less true today because the kernel creeps into a ton more stuff but it's still designed with the same goal in mind. Can you run windows with a modified kernel on a wifi relay device with 99% of the components removed? No; you can run Linux there though.

It's fun to be forced to adapt in producing a Linux/xlib/whatever or OSX/cocoa GUI in some ways because of a process like natural selection: only good code can survive. So it is fun to make sure Xhip works on different platforms because it exposes some of the stupid mistakes I made and forces me to fix them. It's just a heck of a lot of work to do things "the right way".

Is it worthwhile? Doesn't really matter. Much like life itself Xhip isn't worthwhile in any objective sense

It's fun though.

Markku · Post by **Markku** » Fri Jul 14, 2017 3:57 am

Although it's the OS I use and love, would not recommend Linux to musicians. Most computer musicians love their selected toolset. A large part of those tools will not work in Linux. For example, anything like iLok or Pace will not work. There are many VSTs which do work, like all U-He and FabFilter products, but there are limitations.

From a programmer's perspective, I'd say that Linux has quite confusing audio situation. While it's probably most popular in other systems, VST is not even supported by most native Linux apps. Same thing with many other areas, you may find a good solution for audio for a plugin for some toolkit, but UI is left for host to show, which limits the use of any graphics; tools with 100 similar host's buttons may seem quite nerdy...

But. I'm not actually complaining, my system is usable and stable now. I can make music the way I like to do.

Back to Xhip: if 32-bit is dropped, Xhip *may* work on my video PCs, but not in my (32-bit) audio PC. Current version of Xhip is mature enough to use, so I will only lose new features then. I do understand aciddose's view, it's PITA when you need to support many different systems and compile different versions. But I am happy with current state of Xhip.

aciddose · Post by **aciddose** » Fri Jul 14, 2017 11:45 pm

I definitely don't have any plans to drop compiling for 32-bit. The current version though hasn't been tested much by me running in 32-bit mode.

So I was a bit put off to find that nobody had reported that the phaser computes coefficients incorrectly at high frequency settings while running in 32-bit. This is somehow caused by the compiler optimizations as when I disable optimizations the coefficients are correct.

So I wouldn't be surprised if the 32-bit versions have other problems that I'm not aware of. It's easy to compile them though and there is no reason I would stop.

I brought it up because a lot of people seem to click on both at the same time. They start downloading the 32-bit version before their 64-bit download even completes; we're only talking about 1mb which only takes about 30 seconds at the most! I'd put at least 30 seconds into thinking about whether I really needed to click on or download both. Mutant's comment was probably accurate: that a lot of people download both to keep one as a backup.

Bandwidth isn't an issue. Xhip.net does a lot less than 1gb on a monthly basis.

DuX · Post by **DuX** » Sat Jul 15, 2017 11:32 pm

I download both versions because I use both versions.

I have XP as my main OS on one computer and 7x64 as my secondary OS on my second computer, and 64-bit Linux as my Internet OS on both.

Mutant · Post by **Mutant** » Sat Oct 07, 2017 12:24 pm

Beat Magazin tested Xhip on 27.09.2017.
http://www.beat.de/test/freeware-plug-a ... 72362.html

Google translate:
"Xhip for Windows provides powerful digital sound combined with flexible subtractive tone generation. The spectrum ranges from cheap chipsounds to overlapping leads and powerful pads to thick dry basses. Xhips is always in the mix and can be combined very well with warm analog sounds.

This article has been published in our Issue Issue 141."
Also it should be "both oscillators can generate a super saw", but maybe Google failed here.

aciddose · Post by **aciddose** » Sat Oct 07, 2017 4:34 pm

Hard to say, the interpretation of it as "digital" is quite different from comments in the past "it sounds so analog"

The unison feature is still a kludge after so many years, but in the current alpha I'm finally able to reliably hack the synth up and add proper replacement oscillator unison. It isn't surprising that it would be confusing to use because it definitely is confusing.

I'd like to get a few more routing features implemented like random on key-press, "drift" keyed by voice index which will give an unpredictable detuning effect. I've for a long while been wanting to add filter modes to the alpha too although they're 3x as expensive and I don't consider them particularly worthwhile it might be interesting to have them.

quantizer1.5.png

I think I'll be 99% mucking around with Linux and OSX and GUIs though for the next while. Although the skinning feature is coming along nicely and will be useful.

fluffy_little_something · Sat Oct 07, 2017 5:30 pm

Mutant wrote:Beat Magazin tested Xhip on 27.09.2017.
http://www.beat.de/test/freeware-plug-a ... 72362.html

Google translate:
"Xhip for Windows provides powerful digital sound combined with flexible subtractive tone generation. The spectrum ranges from cheap chipsounds to overlapping leads and powerful pads to thick dry basses. Xhips is always in the mix and can be combined very well with warm analog sounds.

This article has been published in our Issue Issue 141."
Also it should be "both oscillators can generate a super saw", but maybe Google failed here.

You have been trying to show on a dozen pages how analog Xhip sounds, and now that verdict...

Seriously though, that review is super shallow, I wouldn't listen to those people...

Mutant · Post by **Mutant** » Sat Oct 07, 2017 7:38 pm

aciddose wrote:The unison feature is still a kludge after so many years, but in the current alpha I'm finally able to reliably hack the synth up and add proper replacement oscillator unison. It isn't surprising that it would be confusing to use because it definitely is confusing.

What is wrong with it ?
I think it is one of the best (only few payware synths rival it) unisons i remember in any VSTi i tried.
I hope the oscillator unison is not going to replace full signal path (excluding effects) unison, but will just add another option.

aciddose wrote: I'd like to get a few more routing features implemented like random on key-press, "drift" keyed by voice index which will give an unpredictable detuning effect.

Nice.

It will free up the 2nd LFO i most often use to emulate drift.

aciddose · Post by **aciddose** » Sat Oct 07, 2017 7:59 pm

Actually a rough emulation of slew rate, rail noise and tuning drift might be an interesting extra feature for the oscillators if I'm going to start adding back-panel pages for oscillator unison anyway.

For example different positive/negative impulses can be used in the oscillator to effect the pulse waveform... although ideally a special "saturated ramp" pulse waveform would be used for that purpose. It may end up being added if I can get xmod to work reliably on my nth order vector oscillator. If so I'll be able to add all sorts of crazy waveforms too like the famous alpha-juno subs and ramp&pulse waveforms. That's technically already working but I haven't bothered to spend the time to fix a bug that makes negative frequencies fail (so xmod doesn't work.)

It would be difficult though no matter what to get real-time frequency modulation "drift" because it's a per-sample computation of the modulation and then the frequency (= xmod), while the route page would just use note-on or low modulation rates... actually that's an interesting idea too although there is technically no destination to modulate oscillator frequency without a proper mod matrix... I think no matter what it ends up being quite expensive to process.

What I was talking about is more just emulating the different state of voices. So each voice would have a random value on note-on that could be mapped to pitch, and that value would change slowly over time but not in real-time. So it would only actually change when pressing the key again. That would give the proper effect though, you know playing C3 and C4 where they're actually slightly detuned depending on which voices are used to play them. That is a really cool effect in analog poly synths and "random" (as in osc unison settings) doesn't quite cover it because the random value changes each time while normally each voice would have a mostly static detune amount instead.

Another interesting thing to play with is adjusting the pitch scale factor so 1 octave = 0.9998 octave. Those are the sort of "version 10.x" "back-panel" features I mention in the TODO.

As much as I'd like to, Xhip really shouldn't be made into a crazy analog emulation

It's mostly just a waste of CPU cycles and not being that is the whole point of Xhip.

Mutant · Post by **Mutant** » Sat Oct 07, 2017 8:14 pm

aciddose wrote:What I was talking about is more just emulating the different state of voices. So each voice would have a random value on note-on that could be mapped to pitch, and that value would change slowly over time but not in real-time. So it would only actually change when pressing the key again. That would give the proper effect though, you know playing C3 and C4 where they're actually slightly detuned depending on which voices are used to play them. That is a really cool effect in analog poly synths and "random" (as in osc unison settings) doesn't quite cover it because the random value changes each time while normally each voice would have a mostly static detune amount instead.

OK

Though i am not sure there would be a big audible difference between the 2 approaches - for sure it would make Xhip behave closer to analog.

Mutant · Post by **Mutant** » Sun Oct 22, 2017 7:46 pm

I figured out settings for Xhip to emulate the power supply effect on the oscillators in Minimoog / The Legend.
Nothing ground breaking or even easily audible, but if Synapse Audio decided to emulate it in their synth, then it probably is important.

Maybe it is that 1% that makes it closer to the 100%.
Here is what i am talking about:

I believe this oscillator flutter is the main reason oscillators on The Legend are that little bit better (if wider is better) than on most if not all other virtual analogs.
OK, so the settings:
Oscillators / A Mod = 0.8 (this controls the height of the 2 little peaks)
Modulator A = Audio rate, Ramp Shape, Width 50%, Keytrack = 0
For 50Hz power source, Modulator A rate = -29st,0cnts (which is 48.999Hz, but close enough)
And for 60Hz power source, Modulator A rate = -25st,50cnts (this controls how far from the main frequency peak the 2 peaks are)

BTW In the picture above, Xhip is green and The Legend is red - i rememember few months ago i confirmed that it looks like that on a real Minimoog too (got a faint DejaVu here... Maybe i wrote about it already somewhere else ?)

Maybe a feature request for Aciddose since you are already working on the drift, which can be considered about the same area (instability) of what makes virtual oscillators sound like analog ones.
Just a simple 3 way switch, OFF, 50Hz, 60Hz and a hidden LFO that does the magic behind the UI - it wouldn't have to be polyphonic because 1 PSU should have exactly the same rate of effect on all oscillators.

[edit]
Just checked and it looks like Diva doesn't emulate this under the hood (of course it is possible to do exactly what i did in Xhip, but that would also require sacrifying 1 LFO).

fluffy_little_something · Sun Oct 22, 2017 8:11 pm

In a nutshell, what are you doing there? Modulating what with what?

Mutant · Post by **Mutant** » Sun Oct 22, 2017 8:16 pm

fluffy_little_something wrote:In a nutshell, what are you doing there? Modulating what with what?

Hmm ?
I'm sorry if it is not obvious from what i wrote above under "OK, so the settings:".
Main pitch modulated by LFO A.
I had to do an experiment to know how much of the modulation to apply and to figure out what semitones and cents under middle C to dial in to achieve 50 and 60 Hz.

Xhip vs Analog Hardware