Does this VST effect exist?

VST, AU, etc. plug-in Virtual Effects discussion
KVRian
847 posts since 20 May, 2010

Post Sun Nov 15, 2015 2:09 pm

I want a VST effect that is able to transform the spectrum of a dynamic spectrum evolving audio file to closely follow the
static average spectral volume amplitude dynamics of the learned reference audio track and adapt to the correct settings over time to tame any jumps in any frequency range that can happen unexpectedly over time.

And making the spectrum sound perfectly high-passed filtered and low-pass filtered as if the instruments was never
transposed/pitch-shifted in semitones/octaves, back and forth.

And please don't destroy all plucks/transients/hits/pizzicatos/staccatos/picks/nails/sticks/hammers... ,etc...
And don't even create audible volume pumping or compression artifacts such as:
clipping/crackless/clicks/saturation/distortion/aliasing.

And use only the volume of the band splits to the control equalizer peaks. Or else phasing artifacts can occur.

Compression artifacts should be removed before they are created by using event declicking and event smoothing
on amplitude volume events before controlling the equalizer peaks.



I hope someone creates a VST effect plugin which does this thing.



These knobs should only be available to the user:

1. Processing amount.
2. Compressor/limiter attack envelope.
3. Compressor/limiter decay envelope.
5. Band splitting steepness/bleeding/crossovers.
6. Dynamic equalizer peaks resonance bandwidth.
4. Event smoothing amount.



And these buttons should only be available to the user:

1. Enable event declicking.
2. Learn spectral dynamics/music frequency response/current average detected equalization.


-----------------------------------------------------------------------------------


The processing amount knob controls both input gain and limiter ceiling at the same to make it work better. And using a equalizer before using this VST effect might make it respond better when it attempts to transform the current frequency spectrum frequency response into the target reference/learned average equalizer band split profile that it tries to match dynamically over time as the problem audio material changes unexpectedly.

User avatar
KVRAF
1789 posts since 9 Apr, 2011

Post Sun Nov 15, 2015 2:14 pm

There are static "matching EQs" but as far as I know no dynamic ones, no.

This is probably a factor: a reference track is not a static frequency range unless it's just, like, one held chord made of sine waves. It, too, is dynamic. I suppose one could take an average of the dynamic spectra like a matching EQ does. I assume you'd want to use a tool like this for mastering?

KVRian

Topic Starter

847 posts since 20 May, 2010

Post Sun Nov 15, 2015 2:17 pm

nineofkings wrote:There are static "matching EQs" but as far as I know no dynamic ones, no.

This is probably a factor: a reference track is not a static frequency range unless it's just, like, one held chord made of sine waves. It, too, is dynamic. I suppose one could take an average of the dynamic spectra like a matching EQ does. I assume you'd want to use a tool like this for mastering?
Does:
Pro Audio DSP Dynamic Spectrum Mapper V2:
http://www.proaudiodsp.com/products/dsm/
Fail?

User avatar
KVRAF
1789 posts since 9 Apr, 2011

Post Sun Nov 15, 2015 2:46 pm

Ah. Nope, that looks like exactly what you want

KVRian

Topic Starter

847 posts since 20 May, 2010

Post Sun Nov 15, 2015 3:08 pm

It's better have several plugins that does the same thing than only having one plugin that does that thing because the same thing can be done in many different ways that make them sound unique to each other.

A equalizer from company XYZ might not sound like the equalizer from company ZXY.
Even if the parameter settings is identical.

User avatar
KVRAF
3751 posts since 25 Jan, 2014 from The End of The World as We Knowit

Post Sun Nov 15, 2015 8:02 pm

Jedinhopy wrote:Pro Audio DSP Dynamic Spectrum Mapper V2:
http://www.proaudiodsp.com/products/dsm/
That is quite a flexible tool, and only $87 to Friday at PA with bx_boom.
buying happiness

KVRian
845 posts since 10 Jan, 2010

Post Tue Nov 17, 2015 6:21 am

well, there are a few other ones like DSM, kind of: melda MAutoDynamicEQ or MSpectralDynamics, or one of the old elevayta ones (thinking FreEq Boy might be the closest: http://www.elevayta.net/product3.htm, 32 bit only). oh yeah, i think reaper's ReaFIR can do it too (but i haven't tried it in a while). most of these will have way more controls than you wanted, but better to have too many than not enough, right? there are others too that I don't remember, that attempt to do the frequency matching, not _that_ many, but a few. hardware-wise, you might look at the fishman aura. but...
Jedinhopy wrote: And making the spectrum sound perfectly high-passed filtered and low-pass filtered as if the instruments was never
transposed/pitch-shifted in semitones/octaves, back and forth.
this isn't likely to get you what you want. sounds like what you want is to take, say, a C note of an instrument, pitch it up to an E or down to an A, and then apply processing to 'fix' the results of the pitch shifting. but on most real instruments, it actually sounds different. heck, you can play the _same_ note on different strings of a guitar, and they'll sound different. playing the C on a piano will resonate differently than an E because of the size of the wood, other strings that sympathetically resonate, etc. if you just pitch-shift up to an E, you may reproduce resonances from C that shouldn't exist at E, while leaving out the ones at E that don't exist at C. and if you're looking to have processing reproduce that... maybe it's possible, but it's going to be way more complicated than an averaging EQ - it has to take a snapshot of what each note should look like, and then every time you play a note, figure out what the note is, load the expected signature, and apply it. And if you're going to go that far, you might as well sample.

existing tools do get you a lot closer - DSM in particular has been really good on my variax acoustic models - but it's still not quite like the real thing. it's still an average, and especially for sampling, it's easier and more accurate just to sample a lot.

User avatar
KVRist
411 posts since 25 Apr, 2007 from Northern CA

Post Tue Nov 17, 2015 6:59 am


KVRian

Topic Starter

847 posts since 20 May, 2010

Post Tue Nov 17, 2015 7:01 am

Now i have installed and tried DSM.

The dynamic EQ matching is not crude enough.
Because highpass/lowpass filter settings from Image Line Harmor can't be captured 100%.
Especially if they use crude filters like Crude low pass + Crude high pass.

KVRian
845 posts since 10 Jan, 2010

Post Tue Nov 17, 2015 7:36 am

Dynamic EQ matching is usually designed for small changes imho - the difference between the input and output spectrum is within a few DB (maybe up to 10 or 20 db). A HP can be much more drastic than that. Maybe MAutoDynamicEQ could do it with some help - you can set a dynamic band to be a shelf or highpass/lowpass where the amount of change depends on the input signal, but it probably won't automatically sample that.

It sounds like really what you are looking for is more along the lines of resynthesis or a vocoder with many bands (reason has one that can do 512 bands, and ReaFIR can kind of act like this too, and some vocoders can 'freeze'), but that act sort of dynamically. maybe something FFT/iFFT related.

one thing is that you basically have an input (the signal you are modifying) and sample (signal you want it to sound like) signal. if you change the spectrum 100% to the sample spectrum, you aren't using the input signal at all (except maybe for amplitude) and might as well use the sample directly. you really want some kind of signal partway between then (normalized).

like i said, the fishman aura sounds like it does something similar, but I don't know of an equivalent VST.

KVRian

Topic Starter

847 posts since 20 May, 2010

Post Fri Nov 20, 2015 2:54 am

(Pro Audio DSP Dynamic Spectrum Mapper V2) creates unwanted volume modulations on the background instruments when trying to mastering a song that contains strong drums on top of all the other instruments that plays in the background at a lower volume.

And i have tried to use zynaptiq unfilter.
But it removes timbre qualities that is important to acoustic instruments such as guitars/choirs/vocals/woodwinds/strings/drums/basses.

And then i try to use zynaptiq unmix::drums to attempt to try taming rhythmic percussive drums to better blend with the background orchestra.

But both of those zynaptiq plugins creates unwanted panning effects in the stereo image.


(Pro Audio DSP Dynamic Spectrum Mapper V2) on the other hand creates volume tremolo effects on the background instruments that is not wanted either.


It's better if the frequency bands does not contain more than one instruments or the other instruments sharing the same frequencies will have their closely spaced harmonics ducking in volume unrealistically.

Return to “Effects”