Old 07-27-2019, 12:58 AM   #1
ManyTracks
Human being with feelings
 
Join Date: Mar 2013
Posts: 356
Default 44.1k sample rate-still allow 75 kHz sounds?

44.1 project sample rate.

Abstract:
How come that in a 44.1 kHz project you can generate tones of 75 kHz



TONE GENERATOR QUESTION and more....
================================================== ==

1.
Please start a project in 44.1 kHz in Reaper

2.
Insert one track with Reaper Tone Generator

3.
Insert Reapers own Spectrum Analyzer

4.
Now set the Tone Generator as follows Basically default with a few exceptions).


Wet Mix: -9 dB
Dry Mix - 6 dB
Base Frequency: 10 000 Hz
Note: A
Octave: 1.0 (Note! Default is "0" so up this from 0 to 1
Fine Tune Frequency: 0

5.
Look at the Spectrum Analyzer. Observe the tone of 20 kHz that is now displayed to the right. This is the 10 000 Hz tone with the "Octave 1.0" making it raised one octave. OK, now we see we have a 20 kHz tone running in our 44.1 kHz project.

6.
Use the slider "Base Frequency" and hold down "Ctrl" while you move the slider otherwise it'll make to big jumps in frequency. Now move the slider carefully to the right and watch the frequency as it approaches 11025 on the Tone Generator (with 1.0 octave this is then 22050 Hz. That's right at the far right end of the Spectrum analyzer. You will see the level is going down there (Nyquist frequency etc), Fine!

7.
Now CONTINUE slowly nove the Vase Frequency sluder towards right. You will now see the frequency reading on the Tone Generator INCREASING. At the same time after a while, you will see the Spectrum analyzer showing a vertical bar indicating your frequency, but that will move towards the LEFT. So as you INCREASE the frequency of Reapers Tone Generator the Spectrum Analyzer is displaying that the audio output now is seeing a LOWER frequency.

8.
Now it's getting more fun: Try to tap in 17000 in the "Base Frequency" in Reapers Tine Generator. You still have "Octave: 1) so 17000 Hz and Octave one should be 34 000 Hz. On the Spectrum Analyzer you see that it is indicating a frequency of around 10 000 Hz.

9.
Tap in 21550 Hz - with Octave 1 that is 43100 Hz. Now you see that the Spectrum Analyzer is indicating almost exactly a 1000 Hz nice sine wave tone!

10.
Tap in 22000 Hz with "Octave 1 this in reality should be 44000 hz) and you have a 100 Hz tone reading on the Spectrum Analyzer.

11.
Nove the"Base Frequency" with Ctrl-slider carefully further up until you reach 22100 Hz on the the Generator. That is then 44200 hz with Oct1. noted how the output and the Spectrum Analyzer first indicated the note was going way down in the sub bass and then move up again to 100 Hz...

12.
Now try to set "Octave" to "4". Tap in frequency 19325 Hz. That should be 77 300 Hz (high frequency in a 44.1 kHz project where anything beyond 22 kHz basically should be ZERO as far as I thought..) Look at the Spectrum Analyzer. You'll hear a frequency of 500 Hz and the Spectrum Analyzer is confirming that.


13.
If you have time and interest, please repeat this test with otherh project sample rate settings. You will see that the resultant artifact tones may be triggered by other frequencies now.

In 96 kHz porject sample rate you now need 72148 Hz in create to generate the 500 Hz artifact tone. In 44.1 kHz project sample rate you needed 77300 Hz.


THOUGHTS

A) Can I assume that Reapers Tone Generator is generating what it is saying it's generating? I think so myself.It seems fine. Does it automatically adjust for a lower sample rate? Think not - then why would it continue to generate smoothly when I move the slider up in the Octave 1 mode...

B) 44.1 kHz is said to offer a bandwidth of approximately max 22 kHz. Half the sample rate with a small deduction of the LP filter to avoid conflict at the Nyquist frequency etc. How can a 44.1 kHz 77 300 Hz from a generator in a 44.1 project create artifacts at 500 Hz?

C) Is the environment WITHIN a 44.1 kHz inside a PC not in all aspects restricted to the 44.1 kHz limitations but acually allows some signal flow / activity?

D) If very high frequencies somehow can exist within a 44.1 kHz project and even create artifact tones folding back downwards into the audible spectrum, what potential consequences are there?

E) In a 44.1 kHz project I can have a tone generator with let's say a 60 kHz sine wave, automating the level of this tone generator and with will indeed affect the compressors etc in that 44.1 kHz project.


What are your thought about all this?


MT
ManyTracks is offline   Reply With Quote
Old 07-27-2019, 01:22 AM   #2
drummerboy
Human being with feelings
 
Join Date: Mar 2017
Location: Switzerland
Posts: 81
Default Aliasing

I think you‘re seeing aliasing, where Frequencies above nyquist are reflected back into the audible spectrum.
https://theproaudiofiles.com/digital-audio-aliasing/
drummerboy is offline   Reply With Quote
Old 07-27-2019, 01:28 AM   #3
Aeolian
Human being with feelings
 
Aeolian's Avatar
 
Join Date: Jun 2010
Location: Somewhere PRO
Posts: 997
Default

Aliasing... it's a thing.

A significant flaw in digital audio, first came to the attention of engineers about 2 minutes after it was invented.

Upper harmonics will exceed 20khz for most instruments.
We can't hear them, and neither will a 44.1khz converter, but the converter will alias those upper frequencies back into the audible range... and that new frequency will be uncorrelated to the original, and will be dissonant.

It's been going on since day dot
Most amp sims alias like no tomorrow.
Synth1 has aliasing out the ass, and it's very very popular.


Either go to 96khz (or 192khz if you want to be dead certain) ... or cast your sail to the wind and just use your ears
__________________
Resistors are futile, you will be simulated ...
Aeolian is offline   Reply With Quote
Old 07-27-2019, 04:04 AM   #4
ManyTracks
Human being with feelings
 
Join Date: Mar 2013
Posts: 356
Default

EXAMPLE OF A REAL PROBLEM WITH THIS
===================================


1.
A Warm Rhodes piano. At a fairly low level. It has a compressor inserted in the chain.

2.
In the mix I now let the sound through from the tone generator tat generates 80 kHz. This triggers the compressor on the stereo out to react and now the Rhodes piano goeas down in level. For no audible reason...

3.
I now select a synth that I know emits lots of harmonics (distrtion if you like) all the way up to 96 kHz. I play a chord with the syntht - fairly warm mellow sound (but ltis of ultrasonic artifacts). Rhodes make the compressor squeeze again...

4.
I HP-filter the synth heavily below 30 kHz. Play the synth. Hear nothing. The Rhodes piano still goes down as a consequence of the synth generating ultrasonic frequencies that affect the cmopressor...


Many plugins generate - sometimes unfortunately so - HF artifacts. I thought before that 44.1 khz would act as a "filter" preventing higher frequencies from entering the system and "doing things...". Apparently this is not the case (?). I get the feeling anything can happen here and that that finding somehow requires some awareness.

Outputs of "suspicious" plugins and synths LP-filtered to make sure they don't add nasty stuff and cause firther aliasing distortion etc? Checkc project in 192 kHz first to identify what tracks are "potentially harmful tracks" ? No way to detect these ultrasonic "garbage data" unnless going to 192 k mode as the spectrum analyzer wont let you see beyond 23 k in 44.1 khz sample rate.

I think this is a real issue and needs to be dealt with.

Just moving upu to 96 kHz sample rate isn't solving this.

192 kHz isn't a solution either.

One problem here is plugins and instruments outputting stuff they should not output.


MT






Let's say a synth is generating loads of distortion / harmonics all the way up to 96 000 Hz...

I thought when operating in a 44.1 kHz project that the overtones - let's say above 23k or 24k would never affect ANYTHING because because it was a 44.1 kHz project... My idea of 44.1 kHz sample rate was that instruments generating "garbage data" at 30 kHz and 50 kHz etc would not actually cause any harm och concesquences. Becaus eI thought a 44.1 kHz project basically couldn't recognize frequencies that high.

What is the scientific explanation here?

Can indeed a 44.1 kHz project still allow a signal generator to emit audio data with frequencies of 60 kHz etc and that audio data will in turn affect the way dynamic plugins work?

Playing a few instruments with KONTAKT that are 44.1 kHz samples in a 44.1 kHz project

Is it only at the rendering stage that there is a cut off above 22 k (approx). While INSIDE the projects anyything can still happen at any frequency?

MT
ManyTracks is offline   Reply With Quote
Old 07-27-2019, 04:06 AM   #5
EvilDragon
Human being with feelings
 
EvilDragon's Avatar
 
Join Date: Jun 2009
Location: Croatia
Posts: 23,338
Default

If you have a sample rate of 44.1k, that means the Nyquist frequency is 22050 Hz. Anything above those frequencies will be reflected DOWN into the audible range, unless the plugin is doing oversampling and/or bandlimiting. In case of oversampling the plugin CAN produce frequencies above Nyquist, but before the signal is sent back to the host, it is downsampled to the host sample rate, in effect eliminating those higher frequencies.

Read up on aliasing, oversampling and bandlimiting.
EvilDragon is offline   Reply With Quote
Old 07-27-2019, 04:16 AM   #6
Xenakios
Human being with feelings
 
Xenakios's Avatar
 
Join Date: Feb 2007
Location: Oulu, Finland
Posts: 7,925
Default

The Reaper JS ToneGenerator just uses a direct sin (or cos, more precisely) function to generate the sine wave and the simplest possible ways to generate the triangle and saw waves too. So they are going to alias. Only the sine wave could be easily handled, the volume could simply be set to zero when the frequency gets too high. The other waveforms would require more complicated programming and I doubt Cockos is going to do that for a simple utility plugin.
__________________
For info on SWS Reaper extension plugin (including Xenakios' previous extension/actions) :
http://www.sws-extension.org/
https://github.com/Jeff0S/sws
--
Xenakios blog (about HourGlass, Paul(X)Stretch and λ) :
http://xenakios.wordpress.com/
Xenakios is offline   Reply With Quote
Old 07-27-2019, 04:19 AM   #7
ManyTracks
Human being with feelings
 
Join Date: Mar 2013
Posts: 356
Default

1.
Set project sample rate at 44.1 kHz

2.
Insert ReaComp in the stereo bus. Set Ration to 4. Set Threshold to let's say -21 dB.

3.
Now start the tone generator inserted in a track which is set for a frequency of 80 kHz and let's say - 6 dB output level.

4.
It will clearly make the compressor kick in and react.


QUESTION:
Is it reacting as a result of a downwards fold back aliasing tone being created? Or is it simply reacting to an 80 kHz sine wave generated by reapers Tone Generator?



MT
ManyTracks is offline   Reply With Quote
Old 07-27-2019, 04:26 AM   #8
Xenakios
Human being with feelings
 
Xenakios's Avatar
 
Join Date: Feb 2007
Location: Oulu, Finland
Posts: 7,925
Default

Quote:
Originally Posted by ManyTracks View Post
1.
Is it reacting as a result of a downwards fold back aliasing tone being created?
It's reacting to the aliased signal coming from the tone generator. There is no actual 80khz in the signal when the project sample rate is 44.1khz. There was no real 80khz signal at any point since the tone generator doesn't do oversampling. And even if it did, the signal would need to be downsampled at the plugin output, resulting in silence or a very quiet residual signal. Reaper doesn't support running plugins at mixed sample rates, as far as Reaper is concerned, everything runs at the project sample rate.
__________________
For info on SWS Reaper extension plugin (including Xenakios' previous extension/actions) :
http://www.sws-extension.org/
https://github.com/Jeff0S/sws
--
Xenakios blog (about HourGlass, Paul(X)Stretch and λ) :
http://xenakios.wordpress.com/

Last edited by Xenakios; 07-27-2019 at 04:33 AM.
Xenakios is offline   Reply With Quote
Old 07-27-2019, 04:44 AM   #9
ManyTracks
Human being with feelings
 
Join Date: Mar 2013
Posts: 356
Default

Hi ED,

"but before the signal is sent back to the host, it is downsampled to the host sample rate, in effect eliminating those higher frequencies."

This would make perfect sense, how can I set up a simple test to demonstrate this?
ManyTracks is offline   Reply With Quote
Old 07-27-2019, 04:50 AM   #10
Xenakios
Human being with feelings
 
Xenakios's Avatar
 
Join Date: Feb 2007
Location: Oulu, Finland
Posts: 7,925
Default

Quote:
Originally Posted by ManyTracks View Post
Hi ED,

"but before the signal is sent back to the host, it is downsampled to the host sample rate, in effect eliminating those higher frequencies."

This would make perfect sense, how can I set up a simple test to demonstrate this?
You would need a tone generator plugin that does oversampling or other tricks to eliminate the frequencies that would alias.
__________________
For info on SWS Reaper extension plugin (including Xenakios' previous extension/actions) :
http://www.sws-extension.org/
https://github.com/Jeff0S/sws
--
Xenakios blog (about HourGlass, Paul(X)Stretch and λ) :
http://xenakios.wordpress.com/
Xenakios is offline   Reply With Quote
Old 07-27-2019, 04:52 AM   #11
ManyTracks
Human being with feelings
 
Join Date: Mar 2013
Posts: 356
Default

1,
OK, so when we hear harsh nasty side effects of harsh nasty plugins in projects that were in 44.1 kHz and suspect artifacts from plugins and indeed can confirm them, it's more likely these are casued by high leveles of already dtosrting stuf hitting the Nyquist magic 22060 real and slightly above causing downwards havoc that can then land in the audible range?

Could I 100% exclude stuff "being alive" in the project that is above let's say 25 kHz?


2.
When playing instruments (sampled in 44.1 kHz with KONTAKT when Reapers sample rate is set to 192 kHz looking at the Spectrum Analyszer now the instruments show like a "fake extension" of bandwidth (compared to the actual sampled wav file, studied in Adobe Audition for instance). Also, there are artifacts created at 30 k and 40 k etc. When played in 96 k or 192 k. SInce these artifacts are not representing true recorded sound, I wonder they come from? Where is the additional "tail" of stuff coming from? KONTAKT. Even when rendering in "perfect" (sound module etc) it's always there. But it isn't there in the orginal instrumnt sample files.

ANother reason to stay in 44.1 kHz perhaps.

Seems like working in 96 k and 192 k s a convenient way of collecting "garbage data"...


MT
ManyTracks is offline   Reply With Quote
Old 07-27-2019, 04:56 AM   #12
preferred.nomenclature
Human being with feelings
 
Join Date: Dec 2014
Posts: 354
Default

Quote:
Originally Posted by ManyTracks View Post
EXAMPLE OF A REAL PROBLEM
===================================

2. In the mix I now let the sound through from the tone generator tat generates 80 kHz.
I would argue that this is not a “real” problem - no sound generator designed for making music is going to alias in a way that creates phantom side effects in a dynamics processor. The real-world problem with aliasing is that it can sound bad in and of itself. If you’re playing a synth and you like the way it sounds, you can stop worrying about it causing problems in your signal chain
preferred.nomenclature is offline   Reply With Quote
Old 07-27-2019, 05:13 AM   #13
ManyTracks
Human being with feelings
 
Join Date: Mar 2013
Posts: 356
Default

I may have been on the wrong path here.

1.
What I've been seeing is well known aliaing foldback downwards artifacts but confused that with the higher frequencies of the tone generator.

2.
If the plugin needs to get entry back into the host with its sound and itindeed needs to do that in 44.1 if the host is set to 44.1, then it appears as if that that would act as a filter of NOT sending back artifacts and nasty stuff up to 80 kHz into the host. It would be prevented from doing so simply becasu it needs to delivere in 44.1 and 44.1 can't deliver 80 kHz music right...?

3.
The problem with plugins creating lots of spikes etc is still real but can indeed be casued by a plugin sending out lots of stuff from 20 and as high up as the 44.1 will allow (sort of 23 k or so right?). That can indeed cause fold back artifacts.


have I got it right you think?


Best thanks for feedback
MT
ManyTracks is offline   Reply With Quote
Old 07-27-2019, 05:27 AM   #14
Xenakios
Human being with feelings
 
Xenakios's Avatar
 
Join Date: Feb 2007
Location: Oulu, Finland
Posts: 7,925
Default

Quote:
Originally Posted by ManyTracks View Post
If the plugin needs to get entry back into the host with its sound and itindeed needs to do that in 44.1 if the host is set to 44.1, then it appears as if that that would act as a filter of NOT sending back artifacts and nasty stuff up to 80 kHz into the host. It would be prevented from doing so simply becasu it needs to delivere in 44.1 and 44.1 can't deliver 80 kHz music right...?
No, there are no magical filters or anything going on. If a plugin causes aliasing, for example by trying to generate a 80khz tone when the host is running at 44.1khz, that's completely allowed, the host can do nothing about it, and the tone will be aliased from that point on in the signal chain. It could be considered a programming error in the plugin, but developers make various trade offs between things when developing. Antialias techniques are typically very CPU intensive and can cause additional latency.
__________________
For info on SWS Reaper extension plugin (including Xenakios' previous extension/actions) :
http://www.sws-extension.org/
https://github.com/Jeff0S/sws
--
Xenakios blog (about HourGlass, Paul(X)Stretch and λ) :
http://xenakios.wordpress.com/
Xenakios is offline   Reply With Quote
Old 07-27-2019, 05:50 AM   #15
Xenakios
Human being with feelings
 
Xenakios's Avatar
 
Join Date: Feb 2007
Location: Oulu, Finland
Posts: 7,925
Default

Quote:
Originally Posted by ManyTracks View Post
When playing instruments (sampled in 44.1 kHz with KONTAKT when Reapers sample rate is set to 192 kHz looking at the Spectrum Analyszer now the instruments show like a "fake extension" of bandwidth
Kontakt is resampling the audio in that case. There are many ways to do that, including ones that add artifacts into the audio. (Noise or aliasing.) Developers may choose to do it in one of those worse ways in order to save on CPU. (Or because they don't consider it a crucial issue.)
__________________
For info on SWS Reaper extension plugin (including Xenakios' previous extension/actions) :
http://www.sws-extension.org/
https://github.com/Jeff0S/sws
--
Xenakios blog (about HourGlass, Paul(X)Stretch and λ) :
http://xenakios.wordpress.com/
Xenakios is offline   Reply With Quote
Old 07-27-2019, 05:52 AM   #16
SonicAxiom
Human being with feelings
 
SonicAxiom's Avatar
 
Join Date: Dec 2012
Location: Germany
Posts: 1,889
Default

Quote:
Originally Posted by ManyTracks View Post
1.
Set project sample rate at 44.1 kHz

2.
Insert ReaComp in the stereo bus. Set Ration to 4. Set Threshold to let's say -21 dB.

3.
Now start the tone generator inserted in a track which is set for a frequency of 80 kHz and let's say - 6 dB output level.

4.
It will clearly make the compressor kick in and react.

...
while you are hinting at an actual technical issue, which is showing on a "microscopic" level, your example is so far away from real life audio engineering that the actual effects can safely be neglected almost 100% of the time. Just listen to a few random CD/radio tracks. Is your listening experience really compromised in any way by aliasing-related artifacts? I'd be very surprised if yes. No relevant sound source generates frequency content @ -6 dB @ 80 kHz!!!

Don't get paranoid! Make music!

.
__________________
[Check out my free VST plugin collection here.]
SonicAxiom is offline   Reply With Quote
Old 07-27-2019, 06:57 AM   #17
ManyTracks
Human being with feelings
 
Join Date: Mar 2013
Posts: 356
Default

"No relevant sound source generates frequency content @ -6 dB @ 80 kHz!!!"


Well, I have seen synths that sound mellow and not particularly edgy ocr harsh but still generate generous amounts of HF energy. And indeed 80 kHz at -6 dB.

In fact many of the major synth brands do this.

If my original suspicion / concern I had about "can 80 kHz sounds interfere while being in Reaper when the project setting is set to 44.1 kHz" well, we need to speak more about that.


I now hear:


A) Plugins in a 44.1 kHz project can never output anything above the traditional bandwidth of 44.1 kHz allows.

B) There may be cases where plugins indeed can sneak in and output stuff all the way "up there" at 80 - 90 kHz...

C) I've seen PLENTY of instruments AND plugins that generate artifacts and harmonics / distortion that reaches all the way up to 96 kHz (that's as high as I can measure in 192 kHz).



So how can we actually test this properly. I thinks a proper good test is warranted here to establish what really is happening.




MT
ManyTracks is offline   Reply With Quote
Old 07-27-2019, 07:10 AM   #18
EvilDragon
Human being with feelings
 
EvilDragon's Avatar
 
Join Date: Jun 2009
Location: Croatia
Posts: 23,338
Default

Quote:
Originally Posted by ManyTracks View Post
In fact many of the major synth brands do this.
Which ones? Digital workstations and synths for sure are not (they all work at 44.1/48k, or in case of Nord and John Bowen Solaris, 96k). Some analog synths can go up to 40k or so but not really higher than that.


By the way, when you're recording a real instrument into your DAW, it doesn't matter if it produces those ultrasonic frequencies. If your DAW project is set to 44.1/48k, this is how analog-to-digital converters on your audio interface will be set up, so all the signals above Nyquist limit (half the sample rate) will be filtered out before recording your signal onto your drive.
EvilDragon is offline   Reply With Quote
Old 07-27-2019, 07:39 AM   #19
SonicAxiom
Human being with feelings
 
SonicAxiom's Avatar
 
Join Date: Dec 2012
Location: Germany
Posts: 1,889
Default

Quote:
Originally Posted by ManyTracks View Post
Well, I have seen synths that sound mellow and not particularly edgy ocr harsh but still generate generous amounts of HF energy. And indeed 80 kHz at -6 dB.

...

MT
so you are stating that some synths obviously produce visually measurable artefacts but you are also stating that you don't notice those artifacts audibly. Well, that's exactly what I'm expecting to happen. Don't worry and carry on with music. I'm not denying at all those aliasing effects actually exist, however, I'm convinced that the audible impact on a mix is WAY too minimal to be of any concern or maybe it even contributes to a better sound. So many other factors do have considerable audible effects that should be of more concern.

Why not discuss the most minuscule sonic effects in an audio forum? Totally legitimate. Maybe discussions lead to them being addressed and finally remedied which would certainly be a good thing. But we should make clear - especially with regard to audio engineering newbees reading this - that the topic has virtually no significance in audio engineering practice.

.
__________________
[Check out my free VST plugin collection here.]
SonicAxiom is offline   Reply With Quote
Old 07-27-2019, 07:44 AM   #20
ManyTracks
Human being with feelings
 
Join Date: Mar 2013
Posts: 356
Default

Hi ED,

ED: "Which ones?"
MT: In the NI KOMPLETE package you will find many, many synths that indeed will generate stuff all the way up to 96 kHz (as far as I can measure). NI's FM8, Absynth, etc etc. Plugin VST synths that are available for everything from "free" to paid plugins for VST AAX etc. There are MANY VST plugin synths I have that are very capable of filling up the entire inaudible spectrum above 20 k...


ED: Digital workstations and synths for sure are not (they all work at 44.1/48k, or in case of Nord and John Bowen Solaris, 96k). Some analog synths can go up to 40k or so but not really higher than that.
MT: Have no experience from that but sounds reasonable that dedicated works stations like that are a bit more "civilized" and controlled.


ED: By the way, when you're recording a real instrument into your DAW, it doesn't matter if it produces those ultrasonic frequencies.
MT: Generally speaking you are correct. If I'm doing a classical recording and using 16 microphones and some are sloping off at 17 kHz and some have spikes at 38 kHz and I'm in 192 kHz then the ultrasonic range will not really represent reality. But you can argue that since you cannot hear it anyway it doesn't matter... And we're back to anither discussion - one is using 96 and 192 as a place to collect unhearable garbage data... If ou're only in 44.1 kHz then of course it's normally not a problem.

ED. If your DAW project is set to 44.1/48k, this is how analog-to-digital converters on your audio interface will be set up, so all the signals above Nyquist limit (half the sample rate) will be filtered out before recording your signal onto your drive.
MT: AD converteras are much better today then lets say 10-15 years ago. While 44.1 may be a great production format, many microphone recordings were done in 96 simply because the AD was so bad and sounded harsh etc. Thi isn't the case today and I suspect much of the 96 / 192 hysteria is based on those old bad AD converters and their aliasing problems.



However - the discussion about certain plugins being able to create really high level HF content 20 k . 96 k WITHIN a 44.1 kHz project is something I'd be happy to investigate further.

If indeed plugin manufactureres can take "shortcuts" and that would enable them to send out artifacts and harmonics of high amplitudes that then can cause fold back aliasing etc even when project sample rate is 44.1 kHz then that's something we would all gain from by investigating and clarifying.

We all want the best audio quality and little things in the background that MAY cause trouble is in our interest to discover and find ways to deal with,

What do you suggest as a next step in finding the best test recipe for this?

Or is the test I just did showing the problem? Are we in fact seeing that it IS possible to have plugins generate frequencies ( as a result from a generator, from artifacts frpm distor
ManyTracks is offline   Reply With Quote
Old 07-27-2019, 07:52 AM   #21
EvilDragon
Human being with feelings
 
EvilDragon's Avatar
 
Join Date: Jun 2009
Location: Croatia
Posts: 23,338
Default

Quote:
Originally Posted by ManyTracks View Post
There are MANY VST plugin synths I have that are very capable of filling up the entire inaudible spectrum above 20 k...
For sure, but ONLY if you run your project in higher sample rates. At 44.1/48k, the inaudible spectrum gets nothing from the plugins, because...

Quote:
Originally Posted by ManyTracks View Post
However - the discussion about certain plugins being able to create really high level HF content 20 k . 96 k WITHIN a 44.1 kHz project is something I'd be happy to investigate further.
This is not the case, though. If you run DAW at 44.1, there is NO WAY that the plugin will generate content above 22050 Hz. It just doesn't happen. What you hear is aliasing (depending on how well or not the plugin combats aliasing artifacts). I thought we have covered that already. Or, in case you're seeing more than hearing, there could also be metering artifacts depending on how a particular spectrometer is set up internally w.r.t FFT, etc.
EvilDragon is offline   Reply With Quote
Old 07-27-2019, 08:04 AM   #22
ManyTracks
Human being with feelings
 
Join Date: Mar 2013
Posts: 356
Default

Hi ED,

(NI's MASSIVE has some great sounds. Run it in 192 kHz and check out the spectrum analyzer. Easy to find many, many instruments generating stuff all the way to 96 KHz. DOn't know why and for whom but it does. But they sound great and it's a wonderful instrumnt. As i U-HE's products. SOme of whic also generate lots of (hopefully) beautiful sounds for birds, dogs and cats). The synth world involves a living species and maybe we need to look at this in new ways...)


Xenakios somehow indicated that it WAS possible, developers taking shortcuts etc. That made me a bit worried. That's why I would welcome a simple but good test recipe...

It makes sense what you say of course. And this is what I have always believed as well. But as always when I think I've discovered something I'd like to test and get the real facts and then move on armed with info I can trust... A test recipe would be fantastic.

I tend to like 44.1/48 simply because it doesn't deal with junk data that no one can hear anyway... Also 20 Hz - 20 kHz is what all EQ's and basically everything is set up for. Moving to 192 kHz and you suddenly have to deal with things beyond the audible spectrum. Like buying a TV set that will show me colors I can't see...

The High End "Hi Res" HiFi audiophiles market has certainly confused many. The older they get and the more money they have the more they believe that paying for pretending to listen to ultrasonic sound is worth while spending time on...

In fact sending lots of ultrasonic stuff to loudspeakers and in and out of things isn't necessarily a very smart ide.

But back to the ultrasonic frequencies within a 44.1 kHz project. I really want to believe you are right but I want to make a test. How can I in front of a colleague PROVE that there is no way that a plugin or synth can generate anything above the Nyquist 22050 region when project sample rate is 44.1 kHz?


;T

Last edited by ManyTracks; 07-27-2019 at 08:10 AM.
ManyTracks is offline   Reply With Quote
Old 07-27-2019, 08:23 AM   #23
Xenakios
Human being with feelings
 
Xenakios's Avatar
 
Join Date: Feb 2007
Location: Oulu, Finland
Posts: 7,925
Default

Quote:
Originally Posted by ManyTracks View Post
How can I in front of a colleague PROVE that there is no way that a plugin or synth can generate anything above the Nyquist 22050 region when project sample rate is 44.1 kHz?
Put a hardware spectrum analyzer to the analog outputs of your audio interface and measure what is going on, I guess. But I am kind of confused now what are you exactly even worried about?
__________________
For info on SWS Reaper extension plugin (including Xenakios' previous extension/actions) :
http://www.sws-extension.org/
https://github.com/Jeff0S/sws
--
Xenakios blog (about HourGlass, Paul(X)Stretch and λ) :
http://xenakios.wordpress.com/
Xenakios is offline   Reply With Quote
Old 07-27-2019, 08:39 AM   #24
ManyTracks
Human being with feelings
 
Join Date: Mar 2013
Posts: 356
Default

I'm curious to know if individual plugins inserted in a 44.1 kHz project are capable of outputting frequencies that are as high as 80 kHz etc.

It is not easy to measure this directly. If I measure that individual track with a spectrum analyzer the project sample rate ( 44.1 kHz ) itself will not measure higher than to around 23 kHz.

SO my concern is that higher frequencies than alittle bit above 22.05 kHz are emotted from individual plugins without me knowing it directly. But that this HF "junk data" may affect other plugins and may of course also be a cause of aliasing distortion.

So How can I design a test that truly and withou doubt verifies if individual plugins are emitting above 23 kHz while sitting in a 44.1 kHz sample rate project?

That is my concern...

MT
ManyTracks is offline   Reply With Quote
Old 07-27-2019, 08:46 AM   #25
ashcat_lt
Human being with feelings
 
Join Date: Dec 2012
Posts: 4,223
Default

While it is obviously related to aliasing, I have trouble really calling it that in the case of the JS Tone Generator because it’s just kind of not really generating the ultrasonic tone to begin with. It’s trying, I guess. It’s calculating individual sample values based on the ultrasonic frequency, but it’s not making enough of them to actually describe that wave. What it actually puts out is actually that lower tone. The ultrasonic tone just doesn’t ever come out of the plugin.

But I guess most of the rest of the aliasing that we do deal with is the same way. Like a distortion that “generates” ultrasonic harmonics. Well, it actually doesn’t. It actually fails to generate those harmonics and what we hear as aliasing is the sound of how it fails.

Without oversampling, there’s nothing in the DAW that can actually make a tone that exceeds nyquist at all. Hardware might, but by the time that comes through your interface, all those higher frequencies are gone. Most of it is just filtered out before it can alias, and the rest will already be aliased by the time it gets to Reaper.
ashcat_lt is offline   Reply With Quote
Old 07-27-2019, 08:52 AM   #26
Xenakios
Human being with feelings
 
Xenakios's Avatar
 
Join Date: Feb 2007
Location: Oulu, Finland
Posts: 7,925
Default

Quote:
Originally Posted by ManyTracks View Post
I'm curious to know if individual plugins inserted in a 44.1 kHz project are capable of outputting frequencies that are as high as 80 kHz etc.
....

So How can I design a test that truly and withou doubt verifies if individual plugins are emitting above 23 kHz while sitting in a 44.1 kHz sample rate project?
They can't emit those frequencies above the limit. Due to programming errors, optimization tradeoffs etc they may end aliasing below the Nyquist limit.

There's no generic test you can do. The plugins either alias to some degree or they don't. (Or have some other anomalies.) If you happen to know what the output signal should be, you can compare the actual output frequencies and their levels to the expected frequencies and levels. Sometimes the aliasing or other anomalies are so blatant you can directly hear them by listening. Developers of course may allow their plugins to alias on purpose too, to get a "crunchy" sound or whatever.
__________________
For info on SWS Reaper extension plugin (including Xenakios' previous extension/actions) :
http://www.sws-extension.org/
https://github.com/Jeff0S/sws
--
Xenakios blog (about HourGlass, Paul(X)Stretch and λ) :
http://xenakios.wordpress.com/
Xenakios is offline   Reply With Quote
Old 07-27-2019, 08:53 AM   #27
ManyTracks
Human being with feelings
 
Join Date: Mar 2013
Posts: 356
Default

ED, regarding VST synths it seems one needs to be a little careful when changing sample rates. Just working on one track here and switching from 44.1 kHz to 192 kHz just "to test" I notice that one NI Massive pad like long chord sounds different in 192 vs 44.1. It's a 4 note chord and one note is a bit louder in the 44.1 vs 192. I don't think it matters in what sample rate you start out, but instruments like a synths that is very "livevly" and moving about all the time seems to be able to behave differently. maybe some synths / presets are a little linked to their sample rate.

Not that this has anything to do with ths "hidden HF-content issue I brought up... Or maybe it has?
ManyTracks is offline   Reply With Quote
Old 07-27-2019, 09:00 AM   #28
ManyTracks
Human being with feelings
 
Join Date: Mar 2013
Posts: 356
Default

Xenakios,

True, that
s what I did with KONTAKT where I hade a known and defined wav-file in the sample library to compare with.

I also made special KONTAKT test instrument with test tones so that I could see exactly what KONTAKT is doing. The result of that test is that if you have an instrument in a sample library that is 44.1 kHz that instrument will only sound 100% good if used in a 44.1 kHz project. ALL other sample rates will degrade the sound more or less. Playing the 44.1 kHz instrument in 48 kHz seems to be one of the worst combinations.

KONTAKT's standard, high and perfect ( in the SOUND MODULE in KONTAKT) are I suppose what's possible to do - Big projects with hunderds or maybe thousands of samples to process on-the-fly isn't an easy task.

Unfortunately the off-line render isn't perfect either once you chose any other sample rate than the original instrument sample rate. Since my libraries tend to be 44.1 kHz most of the, that is the sample rate I go for when using those libraries.

But - you are right - there I have something to compare with. With a plugin that generates stuff it's more difficult.


But would it be POSSIBLE to produce a plugin that generated freauencies of 80 kHz in a 44.1 kHz project? Or is the 44.1 kHz environment automatically by its nature "locking out" everything above approc 23 kHz?
ManyTracks is offline   Reply With Quote
Old 07-27-2019, 09:06 AM   #29
Xenakios
Human being with feelings
 
Xenakios's Avatar
 
Join Date: Feb 2007
Location: Oulu, Finland
Posts: 7,925
Default

Quote:
Originally Posted by ManyTracks View Post
But would it be POSSIBLE to produce a plugin that generated freauencies of 80 kHz in a 44.1 kHz project?
No, it just isn't possible. The Nyquist frequency is a very fundamental thing in digital audio.
__________________
For info on SWS Reaper extension plugin (including Xenakios' previous extension/actions) :
http://www.sws-extension.org/
https://github.com/Jeff0S/sws
--
Xenakios blog (about HourGlass, Paul(X)Stretch and λ) :
http://xenakios.wordpress.com/
Xenakios is offline   Reply With Quote
Old 07-27-2019, 09:54 AM   #30
ManyTracks
Human being with feelings
 
Join Date: Mar 2013
Posts: 356
Default

OK, good. Thanks X. And thanks to ED too.

MT
ManyTracks is offline   Reply With Quote
Old 07-27-2019, 10:05 AM   #31
EvilDragon
Human being with feelings
 
EvilDragon's Avatar
 
Join Date: Jun 2009
Location: Croatia
Posts: 23,338
Default

Quote:
Originally Posted by ManyTracks View Post
Hi ED,

(NI's MASSIVE has some great sounds. Run it in 192 kHz and check out the spectrum analyzer.
I know that! I'm not sure why you're moving goalposts. The thread title talks about 44.1k sample rate, so why would you mention running plugins at 192k and then checking the analyzer? OF COURSE they would generate harmonics above 44.1k's Nyquist frequency - because the project sample rate is now 4 times higher, so Nyquist frequency is also 4 times higher... It's Cpt. Obvious territory.

Quote:
Originally Posted by ManyTracks View Post
ED, regarding VST synths it seems one needs to be a little careful when changing sample rates.
In general a lot of plugins don't like changing sample rates on the fly - you would best reinstantiate the plugin after you change the sample rate.
EvilDragon is offline   Reply With Quote
Old 07-27-2019, 10:23 AM   #32
ManyTracks
Human being with feelings
 
Join Date: Mar 2013
Posts: 356
Default

ED, I'm not a synth expert at all. Although they fascinate me more and more. Mostly dealing with acoustical, orchestral stuff and mostly the goal is to get as clean, natural audio as possible. And that was the initial concern I had. How *plugins* could perhaps generate things / artifacts that even to a tiny tiny minimal degree could affect audio quality was something I wanted to look into.

I was only mentioning synths because they were instruments that so clearly generated audio all the way up to 96 kHz.

Thanks for your great help today and as usual the Reaper forum is a excellent and I do value and appreciate your expertise!

Best
MT
ManyTracks is offline   Reply With Quote
Old 07-27-2019, 11:02 AM   #33
serr
Human being with feelings
 
Join Date: Sep 2010
Posts: 8,275
Default

Quote:
Originally Posted by SonicAxiom View Post
while you are hinting at an actual technical issue, which is showing on a "microscopic" level, your example is so far away from real life audio engineering that the actual effects can safely be neglected almost 100% of the time. Just listen to a few random CD/radio tracks. Is your listening experience really compromised in any way by aliasing-related artifacts? I'd be very surprised if yes. No relevant sound source generates frequency content @ -6 dB @ 80 kHz!!!

Don't get paranoid! Make music!

.
Oh, you mean I won't hear any subtle aliasing artifacts anymore after the CD master has been peak limited and boosted by 14db and then a 20db treble boost? Imagine that!
serr is offline   Reply With Quote
Old 07-27-2019, 03:53 PM   #34
azslow3
Human being with feelings
 
Join Date: Nov 2017
Location: Heidelberg, Germany
Posts: 713
Default

@OP. I suggest you to try understand what is Nyquist frequency and what is aliasing. There are many good explanations and examples.

If you want yet another one, well... here it is

Digital signal is not a "combination of frequencies" you can see on a spectrum analyzer. It is simply a sequence of signal amplitude values. The "project frequency" defines at which distance in time these values are places. So 44.1kHz means the distance between samples is 1/44.1msec. Can you calculate which amplitude 1MHz sin wave will have after 1/44.1msec? Yes, you can. Also for 10MHz, 10HGz, 1Hz or any other frequency. So you calculate and write the next sample, add next 1/44.1msec and calculated the next values. That is how generator you mention works. It was easy so far.

Let say we somehow able to put a speaker cone in position proportional to the sample amplitude value, at correct time (that is what your audio interface + amplifier + monitor try to do). Which sound it is going to produce as the result? How we can describe that sound?
It is logical to assume that if one sample was 0.2 and the next is 0.3, the cone will move "strait". So not moving to 1. and 0. in between. Also physical objects can not move instantly. The sound is transferred by local changes in air pressure.

Our brain is able to preceive "pitch", so the frequency of vibration. Natural vibration has sin/cos form (comes from physical property of materials). So we logically use "sin frequency" to describe sound frequency (not universal for frequency in general, f.e. computer frequency is step function frequency).

What can we take for cone not moving instantly? It has maximum frequency with which it can move, even in case the signal is "true analog 192kHz", you will have hard time to find a cone which can move so fast.

What can we conclude if from 0.2 to 0.3 in 1/44.1msec the cone goes strait? 1MHz sin signal will pass 0.2 and 0.3 amplitude many time during that period. We need to move many times to reproduce it. So there must be some limit in frequency which we are able to reproduce in case we have "move strait" condition...

The rest comes from mathematic pure:
1) the upper frequency limit for our cone is Nyquist frequency.
2) we can represent any function (also in our case discrete samples) as a sum of sin functions. So we can mathematically represent our arbitrary sequence of samples as a "frequency spectrum". That is what you see in your spectrum analysis.

And now conclusions:
I) spectrum of some "simple" function can be confusing at first. Any non continuous function, including step (0, 1, 0, 1), has ALL frequencies component with significant amplitude (1Hz, 1kHz, 10kHz, 10MHz, etc.). That is why they are loved by subtractive synths, you have "all frequencies" from which you can cut something (does not work nice with pure sin wave or fixed set of sin waves as input). There is nothing by itself wrong when something shows arbitrary frequencies in the specter analyzer.

II) in 44.1kHz sampled signal there is NO high frequencies, by the DEFINITION how we interpret it (moving strait from sample to sample)

III) we can sample arbitrary sin waves (f.e. 1.111MHz) at any sample rate (f.e. 44.1kHz). Till we have exact multiple of sampling frequency (88.2, 176.4, etc.) sample values will not be the same. And so there will be some non zero frequency spectrum. Here you will see "aliases".

IV) (bonus) "High-end" audio is at big part about feeding audio signal at frequencies which equipment can not reproduce (remember the speed of the cone). The result is audible artifacts... But if something has "higher frequency" and 100 times more expensive, it should sound better
azslow3 is offline   Reply With Quote
Old 07-30-2019, 04:42 AM   #35
xpander
Human being with feelings
 
xpander's Avatar
 
Join Date: Jun 2007
Location: Terra incognita
Posts: 4,845
Default

Quote:
Originally Posted by ManyTracks View Post
Can indeed a 44.1 kHz project still allow a signal generator to emit audio data with frequencies of 60 kHz etc and that audio data will in turn affect the way dynamic plugins work?
Not 60 kHz signals per se, as said above. But aliasing can mess the control signal paths of dynamic plugins. Here's what Fabien Schivre (Tokyo Dawn Records) says about it in a Gearslutz plugin analysis thread, post #3251 and further...
xpander is offline   Reply With Quote
Old 07-30-2019, 04:52 AM   #36
preferred.nomenclature
Human being with feelings
 
Join Date: Dec 2014
Posts: 354
Default

Read post 3254, he isn’t talking about the kind of aliasing OP is concerned with (although they probably will be now as well, heh).
preferred.nomenclature is offline   Reply With Quote
Old 07-30-2019, 07:40 AM   #37
xpander
Human being with feelings
 
xpander's Avatar
 
Join Date: Jun 2007
Location: Terra incognita
Posts: 4,845
Default

Ah, you're right...if the concern was not how well the plugin itself is handling the aliasing, in this case in the sidechain, my point was moot. I only wanted to say that apparently aliasing does affect dynamic effects. Not having any settings/control over that aspect ourselves as users, it's the plugin choices where we can make the difference, if any.
xpander is offline   Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -7. The time now is 01:58 AM.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2019, vBulletin Solutions Inc.