Announcing beta-test of PKHarmonic VST plugin

[pkane]

Is setting -95,-105,-115,-125 dB or even lower "large nonlinearity" which "introduces large inharmonic IMD components"? Because result is louder. A lot louder.

I think it is because "PKH automatically adjusts the level after each change". Even if it is not needed. And this is probably also one of the reasons why many like it. It is louder.

Let's just back that claim by some measurements, shall we? First, let's record a piece of music (Beatles, Sgt. Pepper) with a large nonlinearity dialed in (around -45dB THD):

Let's measure the RMS level of the track as it's recorded by Audacity:

Now, do the same but with all harmonics set to -200dB:

You tell me if the difference of less than 0.1dB over a track of real music, unweighted, represents "A lot louder" like you claim.

And just to confirm, I matched the two tracks in DeltaWave to measure the true level differences. The spectrum below shows the raw waveforms and the overall level difference between them is 0.0456dB. Is that "a lot louder"?

 

[pkane]

I would photograph my gramophone to use it but that thing is the definition of wire (needle to be correct) with gain,I always wondered what kind of distortion it may have.

I have an old/used TT coming to me in the next few days to play with. Let's see what a TT can do compared to an old gramophone or vs. a modern DAC

 

[audafreak]

Let's just back that claim by some measurements, shall we? First, let's record a piece of music (Beatles, Sgt. Pepper) with a large nonlinearity dialed in (around -45dB THD):

View attachment 293980

Let's measure the RMS level of the track as it's recorded by Audacity:

View attachment 293981

Now, do the same but with all harmonics set to -200dB:
View attachment 293982

View attachment 293984

You tell me if the difference of less than 0.1dB over a track of real music, unweighted, represents "A lot louder" like you claim.

And just to confirm, I matched the two tracks in DeltaWave to measure the true level differences. The spectrum below shows the raw waveforms and the overall level difference between them is 0.0456dB. Is that "a lot louder"?

View attachment 293987

Wrong example.
You compare two PKH outputs with just different settings where both have RMS automatically changed to the same levels. It just says this automatic RMS change is working. Nothing else.

You should compare pure INPUT signal or output with PKH out of the VST chain with PKH changed output. That is what I am talking about.

 

[pkane]

Wrong example.
You compare two PKH outputs with just different settings where both have RMS automatically changed to the same levels. It just says this automatic RMS change is working. Nothing else.

You should compare pure INPUT signal or output with PKH out of the VST chain with PKH changed output. That is what I am talking about.

Since PKH was designed to compare different levels of THD/IMD to each other, the goal was to match output levels while harmonic levels are being adjusted in PKH, and that's what it does. If that doesn't work for you, I'm sure there other plugins you can find that do what you want.

 

[audafreak]

Since PKH was designed to compare different levels of THD/IMD to each other, the goal was to match output levels while harmonic levels are being adjusted in PKH, and that's what it does. If that doesn't work for you, I'm sure there other plugins you can find that do what you want.

I understand. But it is still hard to compare anything as setting all harmonics takes lot of time one by one and during this change you hear many strange sounds. You cannot compare it regularly.
For your intended use there would be better to have possibility to set all levels immediately to -200 by some click, some switch like there and back, manual_set/-200 , on/off. And no strange sounds, mute it during automatic RMS leveling.

 

[pkane]

This is a test... this is only a test.. An updated version of PKHarmonic with a few changes and enhancements (v1.3) is up for beta test. I'm not making this an official release until I get some feedback from a few testers that it works as expected

PKHarmonic64.zip: https://app.box.com/s/b7s2apctg8mxfyz0r5fcvbifetxctd4s
PKHarmonic32.zip: https://app.box.com/s/xyzbdi2xe8bbri9wx5xfradtorrkbogn

Changes include:

• Adding a mixer slider to allow to mix wet and dry signal in any proportion
• Reducing clicks and other noises when making real-time adjustments
• Reducing DC that is generated when even harmonics are specified
• Better matching of wet signal level to bypass level

You may want to take a screen shot of the settings in your current version of PKHarmonic if you have a favorite, just in case it gets reset during the update.

Let me know what you what you find. Here's a screen shot of the new version loaded into Equalizer APO:

 

[audafreak]

One word - GOOD. Now it behaves how it should. Thank you for good work.

I have just one objection about color design. For beginner it is not clear when it works and when bypass. I would let the colors for ON (they are nice) and instead current 1 bypass button create 2 buttons or maybe radios for switching on/bypass just to see immediately what the status is. Now we have to learn what background color means what by trying. And after some time we forget it and have to learn again. It should be clearly displayed.

 

[pkane]

One word - GOOD. Now it behaves how it should. Thank you for good work.

I have just one objection about color design. For beginner it is not clear when it works and when bypass. I would let the colors for ON (they are nice) and instead current 1 bypass button create 2 buttons or maybe radios for switching on/bypass just to see immediately what the status is. Now we have to learn what background color means what by trying. And after some time we forget it and have to learn again. It should be clearly displayed.

Thanks for testing! One way to avoid color confusion with the bypass switch is using the new mixer control: set it to 0% to bypass, 100% to full non-linearity. I should probably set it to 0% automatically when Bypass button is pressed to make this more obvious, too.

 

[Gruesome]

Pkane, v1.3 is definitively an improvement with my particular setup (on which it didn't really work before). It still develops huge (>5 s) delays when making changes (select bypass, or change the percent mixer, or change individual harmonics setting) inside the plugin in the Eq APO editor.exe on my windows 11 surface pro 9 tablet, but it never got to the point where it stopped responding to changes, as it did before.

Interestingly, when I turn the whole plugin on/off within the Eq APO editor.exe, I get an immediate response, and I did not see a buildup of delay there with repeated use of the on/off button.

I also got the impression that the delay between a setting change and the audible effect increased during the first say 30 adjustments/changes, from less than a second to >5 s. There is also still a bit of crackling and volume oscillation after some adjustments. I do not see any significant CPU or memory load while doing this test. It seems as if something is slowing down the plugin so much that a change that should be instantaneous (micro- or milliseconds?) takes seconds while the audio keeps playing until the intended effect settles. There is no antivirus or other special software running (that I am aware of) besides what Microsoft provides with the system (I do see an antimalware executable in the task manager, but it's not taking any significant resource fraction either).

Puzzling, but it now is fun to play with. It's amazing how much distortion one can dial in, depending on the type of music, before it becomes noticeable and then annoying. This in contrast to single tones, where a fraction of a percent is noticeable.

 

[Gruesome]

I’ve tested this recently on an I9 computer with EQ APO. I had no problems. What sample rate are you using for playback?

The default for my headphone adapter, 32 bit 48kHz. (I think for the built-in speakers it's 16 bit 48 kHz).

 

[pkane]

Pkane, v1.3 is definitively an improvement with my particular setup (on which it didn't really work before). It still develops huge (>5 s) delays when making changes (select bypass, or change the percent mixer, or change individual harmonics setting) inside the plugin in the Eq APO editor.exe on my windows 11 surface pro 9 tablet, but it never got to the point where it stopped responding to changes, as it did before.

Interestingly, when I turn the whole plugin on/off within the Eq APO editor.exe, I get an immediate response, and I did not see a buildup of delay there with repeated use of the on/off button.

I also got the impression that the delay between a setting change and the audible effect increased during the first say 30 adjustments/changes, from less than a second to >5 s. There is also still a bit of crackling and volume oscillation after some adjustments. I do not see any significant CPU or memory load while doing this test. It seems as if something is slowing down the plugin so much that a change that should be instantaneous (micro- or milliseconds?) takes seconds while the audio keeps playing until the intended effect settles. There is no antivirus or other special software running (that I am aware of) besides what Microsoft provides with the system (I do see an antimalware executable in the task manager, but it's not taking any significant resource fraction either).

Puzzling, but it now is fun to play with. It's amazing how much distortion one can dial in, depending on the type of music, before it becomes noticeable and then annoying. This in contrast to single tones, where a fraction of a percent is noticeable.

5s delay wait for bypass is likely due to the buffering done by your driver or audio card. Bypass is applied to the audio stream instantaneously, in less than a millisecond, but if the buffer was already filled up with 'unbypassed' samples, it'll take some time to become audible until all the samples in the buffer are output. VST plugin has no control over this. I'd suggest you check buffer size and latency settings in your audio chain.

Changes to harmonic content will take longer. These take about a second or so on my PC. It is computationally intensive and requires double floating point which may be slow on a laptop or another mobile-style processor that doesn't include floating point acceleration. The time to recompute harmonic content will also be decreased with more CPU cores, as this is done in parallel, using multiple threads.

 

[pkane]

The default for my headphone adapter, 32 bit 48kHz. (I think for the built-in speakers it's 16 bit 48 kHz).

Regardless of your audio card, the VST plugin always processes samples as floating (or double floating) point at input/output. The conversion for audio card is done later, by the audio driver or VST host after it receives the adjusted samples.

 

[Sokel]

Did a little test to see actual output and actual CLEARLY audible distortion WITH MUSIC.I tested 3 classical music plays and distortion was more evident during the passages where music was not so dense,like the start of builting a crescendo,The low end also seemed fuller somehow.

Then I captured it with REW just to confirm.
Test is with REW,mic about 30cm far,pointing between about mid and tweeter (but bass is also a little visible).
Player is foobar with no other DSP loaded,playing the sweep tones generated with REW (appropriate IR file also loaded in REW)

So to reach the above audible under any circumstances result configuration ended like this:

We're talking distortion way above the 0.1% industry standard which is a bittersweet result.

The output looks like this (AC was on on silent mode but still does it's thing on highs over 5-6Khz,you can tell by the noise foor,so ignore):

without added harmonics percentage

with added harmonics percentage

without added harmonics SPL

with added harmonics SLP

Nothing concrete about it of course,nevertheless it was fun!

 

[maty]

I like the improvement!

Tested with JRiver Media Center 31.0.38 64 bits

Chrome browser based -> WDM driver

WDM Driver - wiki.jriver.com

wiki.jriver.com

[ JRiver Media Center version 20 for Windows introduced a WDM Driver (a virtual audio driver), which can be used to re-route all windows sound through Media Center ]

Greatest Audiophile 2023 - Best Voices & Instruments - Audiophile NBR Music

 

[maty]

With audio files. Better with my old convolution, without flat acoustic phase.

H2, H4, H6 and H8

 

[maty]

After many tests, for commercial music recordings with real instruments, the best combination for my second audio system:

With classical and films, better H2 at -80 dB. I have to do more tests (classical recordings) because I have focused on Hotel California, a theme that I always use to verify the sound changes.

 

[Gruesome]

Did a little test to see actual output and actual CLEARLY audible distortion WITH MUSIC.I tested 3 classical music plays and distortion was more evident during the passages where music was not so dense,like the start of builting a crescendo,The low end also seemed fuller somehow.

Then I captured it with REW just to confirm.
Test is with REW,mic about 30cm far,pointing between about mid and tweeter (but bass is also a little visible).
Player is foobar with no other DSP loaded,playing the sweep tones generated with REW (appropriate IR file also loaded in REW)

So to reach the above audible under any circumstances result configuration ended like this:

View attachment 301994

We're talking distortion way above the 0.1% industry standard which is a bittersweet result.

The output looks like this (AC was on on silent mode but still does it's thing on highs over 5-6Khz,you can tell by the noise foor,so ignore):

View attachment 301996

without added harmonics percentage



View attachment 301997

with added harmonics percentage




View attachment 301998

without added harmonics SPL


View attachment 301999

with added harmonics SLP

Nothing concrete about it of course,nevertheless it was fun!

It looks like you reversed 'with' and 'without'. The 'without' graphs have higher harmonics.

 

[Sokel]

It looks like you reversed 'with' and 'without'. The 'without' graphs have higher harmonics.

???
The description of each photo is under the photo,I don't see higher.

 

[Gruesome]

???
The description of each photo is under the photo,I don't see higher.

without : THD_H2..9 3.41%
with : THD 2.37%
without : THD 3.39%
with : THD 2.22%

 

[Sokel]

without : THD_H2..9 3.41%
with : THD 2.37%
without : THD 3.39%
with : THD 2.22%

Oh,you mean the freq where the cursor is shown at the results below.
It's on 26.2Hz (not for any reason) while my speakers have F3 at 31Hz so nothing to see down there.

Furthermore,measurement is 30cm from the midrange-tweeter,like nearfield them,woofer is way down from there,you can tell by the response.
I should have gone more than one - two meters to get all of the drivers and then I would have the room there too.So...

Look higher,the red line.