Wavestation v1 vs v2

[whitesun]

Same preset doesnot null on v1.x and v2.x. Desciption of v2 mentions some revising the algorithms. What's opinion of hardware owners of ws? Which version is closer to original hardware as to sound? Indeed you hear the differences even without nulling tests.

[muzines]

Can you give a specific example?

Many synth sounds won't null because of free-running modulations and FX etc - as far as I can see, the engine has not changed between versions, just the interface.

The software versions sound identical to me, and neither sound the same as the hardware, as the hardware has a more lo-fi, grungy, "defocused" sound due mostly to the D/A converters of the time. The software versions therefore sound more "hi-fi", clearer, "better", and need a little "roughing up" to sound closer to the hardware. But the core waveforms, FX and features are identical, so the sounds will be the same, just more "polished" in the software versions.

[whitesun]

V1 is softer and more cloudy with less harmonics in highs with low mids boosted. V2 is crisper but more dirty like overdrive was added. I don't think modulation can make the same difference in sound on majority of presets. You hear better each oscillator in v2, but with v1 it kind of better makes one whole of all of oscillators.

[muzines]

No, if they sounded different it would break project compatibility, and I've not noticed any particular difference.

When you say "Desciption of v2 mentions some revising the algorithms" can you link to what you are referring to?

All Korg says on their page is:

The KORG Collection WAVESTATION V2 is an upgrade of the WAVESTATION software from 2004 and has been refined for modern production environments. We added the much requested high-resolution and scalable interface. The preset sounds include all 1,500 programs, including those from the expansion ROM cards, giving the famous WAVESTATION sounds through the latest software as a plug-in.

https://www.korg.com/uk/products/softwa ... vestation/

...again, making clear that the V2 plugin is mostly about retooling the interface, it doesn't mention any sound engine changes at all.

And from the update notes for V2.00:

Changes in Version 2.0.0

We overhauled the user-interface:
- Supports the UI size scaling. That is available from Utility button in the top screen.
- A brand new sound list view to let you find sound more easily. We categorized the Performances with the tags like M1 plugin.

More sound capability
- Added a randomize button to let you find a sound in the different way."

https://support.korguser.net/hc/en-us/a ... AVESTATION

So again, no changes to the sound engine in the update notes for V2 at all, and none in any updates since 2.00 to the current 2.4.2 version either.

[whitesun]

I hope korg will answer the question. I am not deaf and differences are clear! Perhaps i have misunderstood something in description as to 'revising'. I did not store the link or maybe it concerned iwavestation for ipad? I hezitated about choosing between v1 vs v2. V2 sounds more appealing as it's crisper for me but after reading all comments on all forums in internet about hardware vs software versions i have come to my personal conclusion that v1 was closer to real hardware than v2. Nulling tests can be made - you just have to switch off all modulation and compare same presets. Perhaps i will do it since no one but me hears the difference.

[whitesun]

They write.about v2: For the hardware, KORG analyzed the circuit diagrams of the original design, and completely replicated the circuitry in software down to the last detail. Engineers with an exhaustive knowledge of the WAVESTATION made careful adjustments in order to obtain the sound of the original instrument, something that cannot be obtained simply by sampling the original sounds.
The above was missing with v1. Does it mean that they just used ws code from hw for v1 but later they added distortion emulation for v2?

[muzines]

I am not deaf and differences are clear!

I've noticed no differences, but haven't directly compared (though I can do so). However, I'm *extremely* familiar with the hardware, and the differences between the hardware and software (as mentioned above) are quite clear.

i have come to my personal conclusion that v1 was closer to real hardware than v2.

Having owned a real WS since the early nineties, programmed it to death, and had the plugin since it was released, I can say that V2 of the plugin sounds no closer to the hardware than V1 did.

They write.about v2: For the hardware, KORG analyzed the circuit diagrams of the original design, and completely replicated the circuitry in software down to the last detail. Engineers with an exhaustive knowledge of the WAVESTATION made careful adjustments in order to obtain the sound of the original instrument, something that cannot be obtained simply by sampling the original sounds.

Yes, that's how they made the Wavestation plugin. All that applies to V1 as well, which was what the engine was written for. And as we have seen from their release notes, there have been zero audio engine changes documented for V2. Not even a "We looked again at our models of V1 and improved them to more closely match the hardware" - *zero* changes.

The above was missing with v1. Does it mean that they just used ws code from hw for v1 but later they added distortion emulation for v2?

You haven't proved that their documented changes are inaccurate, and it fact that they did some audio engine improvements, and break old project compatibility, but didn't want to document this for some reason.

If you want to pursue this, do some tests between both versions, and let me know specifically what tests you are doing, and I will replicate them here. I'm pretty sure there will be no meaningful differences in the audio output between both versions, and this will match up to the documented change log, but I'm happy to explore this, to see if I can hear, and verify, any audio differences between them when examined.

[Lostgallifreyan]

Same preset doesnot null on v1.x and v2.x.

What means 'null'? Do you mean inverse mix of two supposedly identical sounds, looking for residual signal?

Or do you mean a tendency to fail to become entirely silent after playing several notes, as the original DX7 would show, especially on sounds with high mod index?

If the first, don't worry about it. Instead, set up both samples in an ABX test. If you can't hear the difference, you're ok.

If the second, that's down to using logarithms (for fast EG and mod scaling by additions), because with those there is a 'full level', but no such thing as true silence, once converted back to linear representation.

[whitesun]

I used inversion of sound from v1 mixed with non-inversed sound from v2 - the regular nulling test. Indeed i should have turned off all modulation as lfos are usually in nonsync mode. But just fast switching between was enough to hear that some high pitched oscillators were clearly heard with v2 and drowned or masked in v1 by other oscillators and general impression is like v1 uses more sine waves but v2 more saw waves. It can come from shape of crossfades between waves. If it is sine like it's softer. Reminds of granular synthesis. V1 is more airy to me. V2 is more dirty but some may think perhaps that this dirtiness makes it closer to hw. V1 is more defocused and with less 'presence' and that presence is usually from raised 5000per.
I am lazy to do null tests with mod off as i need to reinstall v2 again. May be someone else wants to do them? Besides v2 gui opens with large delay and cpu consumption boost to 28%,not to mention that reaper's presets for v2 do not work with v1 which means no downgrade compatibility, which put scales' cup into direction of v1 though doubtless that v2 has some attractions but those are not that crucial for me to upgrade, not to mention that no one has ever confirmed that v2 is closer to hardware.

[whitesun]

I would also say that v2 has some cunning hidden exciter like in harrison mixbus daw but v1 is more 'antialiased' (as antialiasing filter on camera matrix acts on image) and like processed by wide reverb with later narrowing by dualpan i.e. some washing out details takes place. If compared to abundance of crisp synths proposed to the public v1 can be regarded even as a virtue. I generally don't like that ffted soft sound from 00's era but i don't.think that v2 is closer to the sound from 90s.

[whitesun]

Some advice: if you want to better hear the differences between versions process sound via same reverb instead of comparing raw sounds since as reverb does multiplications it also enlarges the differences.

[muzines]

I used inversion of sound from v1 mixed with non-inversed sound from v2 - the regular nulling test.

If you're entirely basing this V2 sounds different theory from this method, it wouldn't be the first time someone takes the wrong outcome from a flawed null test, unless you are *sure* that you've ruled out all the time invariant/random modulations that might be going on.

Honestly, ABX testing is the proper tool for this - if you can set up blind tests of the same audio exported from each version of the plugin, and reliably, *consistently* identify them, then you're well on the way to demonstrating there are discernible differences. Null tests are easy to do wrong, and hard to do right (I'm not saying you're doing it wrong, but I'm saying there is no supporting evidence other than your null tests that there are differences between the versions, so I'm much more likely to think that the problem lies in the testing approach.)

Besides v2 gui opens with large delay and cpu consumption boost to 28%,not to mention that reaper's presets for v2 do not work with v1 which means no downgrade compatibility

Well this seems strange - I don't use Reaper, but DAWs typically save the state of the plugin, and I can categorically demonstrate that my Logic AU presets - many thousands of them - for the V1 Wavestation recall and work just fine in the V2 plugin, indicating that there are no changes to the plugin state at all.

Honestly, there's nothing I've seen so far that would lead me to believe there are *any* noticeable differences at all between versions in terms of the sound engine - but please do post up some audio clips of the same audio rendered from both versions if you are clearly hearing a difference, and I'll look into it.

I would also say that v2 has some cunning hidden exciter like in harrison mixbus daw

You would clearly see this on an analyser if this was the case. Have you done this?

but v1 is more 'antialiased' (as antialiasing filter on camera matrix acts on image)

If I'm understanding what you are saying here correctly, that's not how digital audio works, there are no correlations to pixellated images. All sampled audio is played back through reconstruction filters that reconstruct the smooth waveform perfectly (within Nyquist).

Some advice: if you want to better hear the differences between versions process sound via same reverb instead of comparing raw sounds since as reverb does multiplications it also enlarges the differences.

But hopefully not as part of a null test procedure for hopefully obvious reasons... Honestly, adding in extra processing is not going to help when you are trying to hear the differences between the two sources. And if you can't reliably hear the differences without adding extra stuff, I'd argue that there probably aren't differences to hear..!

[whitesun]

I am not going to prove anything to anyone, everyone is welcomed to do own testing. You have misunderstood 90% of my info.
1. Presets made in v2 do not work in v1, not vice versa.
2. Re pixels: demagogy! Everyone knows what is a smoothed thing.
3. Reverb won't spoil null tests if you use dualmono reverb. It helps to hear differences as it extends in time resonances and sums them with own resonances. The version without them will just be diffused. Reverb will help in cases where you can't swith off modulation and have to compare by listening.
I will stick.to v1 just to be on the safe side...

[Lostgallifreyan]

Reverb will help in cases where you can't swith off modulation and have to compare by listening.

Reverb is definitely useful, but not for continuous granularity like low-res digital sound (it will smooth it out too much). It can reveal a lot of detail in sudden transients. Either way, whatever it reveals, it is still best to use ABX to compare the 'magnified' results. Better yet, try the sound unmodified, but listen on in-ear phones like Etymotic ER-4s, during the ABX test, because you'll hear differences you'd never notice elsewhere. Bear in mind, that if it takes this rigour to catch them out, you're well beyond diminishing returns in almost any other situation where those sounds are used.

If you're hearing differences in sound during modulations, it can mean there are different internal resolutions in the plugin, and calculation at high speed can cause differences that might not be in the original static array of samples. Without knowing the details of architecture, all bets are off, you can only go by what you hear. A faster approach might be to figure out what masks the unwanted aspects of a sound. Case in point: Mellotron sounds can be awful till they go through a reverb, then they can be awesome. The same is usually true for a violin...

In short, sometimes treatment can be faster than diagnosis, and still get a good outcome.

[whitesun]

I disagree about reverb's usefullness only for transients. It averages and smoothes sound. Ear cannot even detect pitch at length shorter than xy ms (read audio expert book for exact numbers as i don't remember). So, for short granular like sound reverb is especially useful as it was in this case. Without reverb it's hard to impossible to detect differences between raw waveforms. For this case blind testing is not even needed as v2's sound is way more high frequenced and rough which makes v1 closer to hardware. Someone on rmmedia.ru forum claimed even difference between 1.0 and 1.5 with which i also agree but it was very close unlike vs v2.

[Lostgallifreyan]

Continuous granularity (like the friction-based noise in a violin sound) is smoothed by reverb. A one-shot short granular sound is a transient, so the reverb reveals more about it because it spreads to a longer time scale, gives ears and brains more to latch on to.

You still need ABX because anything else is too subjective. There may never be a true 'objective' (quantum physics seems to indicate it's impossible) but ABX is as close as you can easily get. If you really want to compare results with those from other people, you're going to have to do this, otherwise it's like debating whether the metal a mains cable is made of makes a difference to hi-fi. Exotic descriptions can spiral outwards, forever, and you need to work inwards, toward a method that leaves as little as possible to chance. Hence, ABX. It reduces to 50/50 odds. It's crude, but that's what makes it important if you need other people to help you make a critical judgement.

EDIT:
While there are dedicated ABX test programs, you don't need one. What I do is load up two sounds in Sound Forge, put one hand ready over the space bar, and the other over Ctrl+Tab. I use a fast keyboard repeat rate. I close my eyes and hold Ctrl+Tab for a few seconds to scramble the two client windows so I don't know which is selected and has focus. I use Space to play (and stop if I want to stop early), then toggle with Ctrl+Tab ONCE to hear the other one. Alternate between them to retry if needed. So long as you don't know which is selected to play, this is a good method. If I'm already working with Sound Forge, it's also extremely convenient. I can even trim the start and end before testing, which an ABX program can't usually do. It's so easy to do it all in Sound Forge, so that's what I do..

[whitesun]

I know this method of blind switching between playbacks - no need to try to present me as a pupil of the first form, and certainly i have done it also while comparing. This method however just states differences without explaining them. Re reverb again: it doesnot matter what the difference is: transients or pitch or anything else: any differences will be exagirated and heardable. It works simply like that: if there are differences they will be revealed. To my mind they could change shape of crossfades from sine to linear because it makes sound rougher and closer to square waveforms, though i really don't care as i have already made my choice. V2 is not worse sounding, it's just further from hw I would say. Any if will produce other psychodelic effect and it's that effect that really matters.

[Tomás Mulcahy]

Has AI taken over?

[whitesun]

These ****** forums have become totally useless since about 2005. It's better to read books than debate with the trolls! I recommend forums which discuss audio stuff from 90s if you can find any. Cheers!

[zenguitar]

The previous post has been reported and a warning has been issued.

Andy

[adrian_k]

Trolls? Kettle meet pot.

[whitesun]

It's bad that they do not allow importing own waveforms and wavetables into the vst. Otherwise it would be ultimate synth. Also it's impossible to run v2 and v1 simultanously. V2 has randomizing feauture but v1 has softer sound and non-freezing gui which makes it the right choice. No one else makes such multilayered wavetable synths.

[muzines]

It's bad that they do not allow importing own waveforms and wavetables into the vst. Otherwise it would be ultimate synth.

You can do that in Wavestate, which is Korg's modern take on the Wavestation (and contains many of the OG Wavestation waveforms).

Also it's impossible to run v2 and v1 simultanously.

Correct, as it's the same audio engine and sounds identical to V1, Korg kept the plugin ID's the same so the one with the updated interface replaces the old one. Projects aren't affected, as they sound the same, otherwise Korg would have given the new version a different ID and leave your projects that used V1 alone.

V2 has randomizing feauture but v1 has softer sound and non-freezing gui which makes it the right choice.

Great, use the version you prefer.

[whitesun]

It doesnot sound the same. Why do you lie, i wonder? Paid by korg? Not a decent one! I hope some masons to program a multilayered clone of their wavestation to load own waveforms and wavetables. Too many restrictions with wavestation but is it they have patented multilayered design for ages?

[Eddy Deegan]

It doesnot sound the same. Why do you lie, i wonder? Paid by korg? Not a decent one! I hope some masons to program a multilayered clone of their wavestation to load own waveforms and wavetables. Too many restrictions with wavestation but is it they have patented multilayered design for ages?

Inflammatory/accusatory posts like this, in combination with others from Whitesun recently, are not welcome here.

A warning has been given and Whitesun is now banned for a week. Should there be a repeat of this kind of nonsense after the temporary ban has expired a permanent one will be issued.