Frequency response vst analyzer

[JohnD71651]

Hi,

Some of my plugins are 64 bit VST3s and I'd like to measure what they are doing. I'm using an oscillator and a frequency analyzer but am looking for other ways of measuring plugins. Is there something like Christian Budde's VST plugin analyzer that can handle 64 bit plugins, or even better, an equivalent 64 bit plugin?

[Compyfox]

Since nobody answered you yet:

As of this moment, there is no real "all in one" standalone tool or plugin test suite (with suitable test signals and analysis modules) that is not only capable of using (native) x64 plugins, but also VST3 plugins. Your only chance to test in VST Plugin Analyzer, is to use some sort of wrapper (i.e. modular environment).


There are ways to do that in your host, however that unfortunately introduces higher tolerances and room for interpretation - depending on what source signal you use, and how "detailed" your analysis tool is. As you mentioned already, a suitable frequency analyzer is a good start. Plus points if you can set up the resolution, the "slope" and maybe even the domain of the analyzer (from Magnitude to Time or Phase).


Then you need 3 essential test tones:
- a sine wave at -18dBFS at either 997Hz or 1kHz (the former shows possible aliasing better) for testing the harmonic distortion. A level of -18dBFS due to the fact that a lot(!) of plugins use this as reference level/hotspot (mostly analog emulation type). Go higher, and you see distortions that shouldn't be there. I do recommend a WAV file of a test tone, not i.e. Cubase's built in test tone. I realized after years of testing and using this plugin, that it introduces offsets. There are plenty of sources for test files on the web at this point.

- a DIRAC test "pulse", which is for measuring the frequency response (the more advanced way)

- a pink noise test tone, which is an alternative to the DIRAC pulse method (less detailed since the readout is "more fluctuating", needs adjustment of the FFT slope)


WIth this, you can already get far and you can measure the two most important things:
frequency response and total harmonic distortion



Sadly, there are not really any suitable tools to read out transfer functions (i.e. static characteristics of a compressor), since you need a measurement tool that can read out the input/output over a specific time frame. You can however measure the attack/release behavior of compressors if you set up a sine test tone sequence at 1kHz and varying signal strengths over a course of (let's say) 5-10 seconds. Then simply run a compressor through at a threshold slightly lower than your medium signal signal strengths in this sequence, and you can read out how the attack/release behavior is, according to the visual representation of your hosts' WAV form readout. This is especially great for understanding program dependent attack/release.


I think, delay times of (i.e.) a delay can be measured with a single cycle sine wave, and then you only measure the "samples" and/or read out the millisecond offset that is shown in your host. But I've never done that within Cubase. And to be honest, I barely needed to test correct delay values.



Unfortunately, I can't offer you a solution to correctly "measure" the latency of a plugin. And this is only scratching the surface. But now you've got a starting point.

[sergiofrias]

Hello ,i don't advise using a Dirac impulse in most cases since if there is some kind of saturation involved in the plugin the frequency response will be very diferent,but if the plugin is free of saturation/distortion then a dirac will be the best choice ,and will display a very fast and clean measurement...i also advise to take measurements with white noise ,since it's most flat on all frequency spectrum .i use voxengo span on "average" mode and it works great (just have to wait a few seconds),don't forget to set the slope to 0...
delay times can be measured with a sine wave and a osciloscope ,jscope vst is also free and is perfect ,since it gives times in ms ,frequency and pitch.
mda test tone is wonderful to testing and also free (white noise,pink noise, dirac"impulse",sine ,sweep)
for transfer functions your best choice is using vst analizer (measurement---dynamics---static characteristic)
or you can try my manual method ,use microsoft excel ,measure a sine wave from - 60 db to 0 db at small increments ,and use a db meter (i use the free rms buddy to measure the peak) place those numbers on the sheet (input against output) and make a x/y graph.

[JohnD71651]

Thanks a lot! That was very useful info by both of you!

[Compyfox]

Hello ,i don't advise using a Dirac impulse in most cases since if there is some kind of saturation involved in the plugin the frequency response will be very diferent,but if the plugin is free of saturation/distortion then a dirac will be the best choice ,and will display a very fast and clean measurement...

Sorry but could you elaborate on that?

To my knowledge, VST Plugin Analyzer uses DIRAC as measurement for the frequency response and not a White Noise or Pink Noise test signal.

i also advise to take measurements with white noise ,since it's most flat on all frequency spectrum .i use voxengo span on "average" mode and it works great (just have to wait a few seconds),don't forget to set the slope to 0...

The main differences are, that White Noise has equal energy per Hz, while Pink Noise has Equal Energy in all octaves (which is better for measuring rooms, it's also used to measure microphones btw). If you use White noise, you might not need a slope adjustment. If you use pink noise, you need a 3dB slope adjustment.

Whatever is better - I don't know. Though Pink is just more common for testing gear.

delay times can be measured with a sine wave and a osciloscope ,jscope vst is also free and is perfect ,since it gives times in ms ,frequency and pitch.

I never looked at Oscilloscope's for measuring delay times to be honest. Then again, JSCOPE is WIndows and 32bit only.

I lost track of Oscilloscopes, especially free ones, but if you have a working plugin bridge, then:
http://bedroomproducersblog.com/2011/07 ... u-plugins/

mda test tone is wonderful to testing and also free (white noise,pink noise, dirac"impulse",sine ,sweep)

MDA Testtone is indeed a good starting point. But I still recommend WAV files over any plugin. Search keys like "test tones" and "WAV" are great sources.

for transfer functions your best choice is using vst analizer (measurement---dynamics---static characteristic) or you can try my manual method ,use microsoft excel ,measure a sine wave from - 60 db to 0 db at small increments ,and use a db meter (i use the free rms buddy to measure the peak) place those numbers on the sheet (input against output) and make a x/y graph.

Interesting aspect that I've never thought of - thanks for that.

So you basically create a test sequence (a sine sweep) of (let's say) 1 minute, with 1dB inrements of a sine wave from -60dBFS to 0dBFS every second, or extend the sequence to 2minutes and do 0,5dB increments every second - which is easier to note down on an Excel sheet. Go faster (i.e. a sine sweep of 5s from -60dBFS to 0dBFS) if you only want to look at the "WAV readout" of the host.

[sergiofrias]

Hello ,i don't advise using a Dirac impulse in most cases since if there is some kind of saturation involved in the plugin the frequency response will be very diferent,but if the plugin is free of saturation/distortion then a dirac will be the best choice ,and will display a very fast and clean measurement...

Sorry but could you elaborate on that?

To my knowledge, VST Plugin Analyzer uses DIRAC as measurement for the frequency response and not a White Noise or Pink Noise test signal.



yes you are correct ,vst plugin analizer does use dirac impulse...if you are using a plugin with high distortion /saturation ,the frequency response will be incorrect ...so you must use white noise instead...



the image above is using the same plugin (waves tg12345 ) with the same settings ,im using a little drive (saturation)...the image on the left is with noise(span) ,and on the right dirac(vst analizer)

please try it

[Compyfox]

I don't have Waves plugins at hand - but you're implying that Christian Budde doesn't know what he's doing, if he's in fact the forerunner if creating plugin measurement tools?!

And of course will distortion influence the frequency response. Unless you can measure "just" the frequency response without the distortion module (which is mostly not the case). So the offset is (IMO!) actually with the White Noise rather than the DIRAC.

[sergiofrias]

I don't have Waves plugins at hand - but you're implying that Christian Budde doesn't know what he's doing, if he's in fact the forerunner if creating plugin measurement tools?!

nothing to do with vst analizer ,i can use a dirac with span and display the same results...

[Compyfox]

Which results are we talking about again? An "offset" or a crazy frequency curve?



EDIT:
I just tested SlickEQ GE (Gentlemen Edition) with one of the saturation modes (Excited and Toast), since this has a built in Magnitude Plot. VST Plugin Analyzer showed an offset, while the EQ built in plot remained flat.

I contacted the developers and try to get an answer what type of "test tone" they are using, or actually recommend. During tests with VST Plugin Analyzer and the "signal generators", I also realized that Pink Noise gives the best frequency representation compared to a White Noise (which only "bows" the signal), but unfortunately the C.Budde Analzer doesn't offer smoothing of the frequency curve or lower the response time (to get a more smoothed readout).

But yes... I can agree in this case, that a DIRAC is not ideal if you utilize saturation and want to get a frequency response measurement. Then again, don't you want a frequency response from the plugin at "default settings" most of the time and measure the Harmonic Distortion separately?

Unless you want to know what type of influence a saturator has on the frequency curve. Though with all the plugins I tested in recent years, that influence is minimal if the saturation is subtle - unless we talk about a fuzz box.

[Funkybot's Evil Twin]

It would be great to see a company take this on. I'd love something like Christian Budde's analyzer, only 64bit, stable, with a few more options, and a bit easier to use (at least with things like attack/release times). Oh, and stable. That thing crashes a lot.

This seems like the kind of thing that Melda Productions, Bluecat, or Voxengo might excel at. Quick, someone point them to this thread.

[Twrogstudio]

or Mr Gamble

[Funkybot's Evil Twin]

or Mr Gamble

I thought about that but feared Dave would charge too much for me to be interested. I was thinking something in the free to $50 range. I'm sure Dave could do an amazing job, but we'd probably be looking at premium pricing.

[Compyfox]

I still have hopes for VST Plugin Analyzer v2. But there weren't any lifesigns of Christian Budde in recent months.

Maybe there will be a tool in the near future - who knows.
So far there is no sparrow on the roof leaking some news...