Invisible Filters ?

Some others strange and intersting facts from Windows XP task manager:
playing module “test2.rns” in background mode:

cpu load: memory load:
1.20 45-55% 9412kb
1.25 50-60% 9948kb
1.5RC2 55-65% 24500kb

The difference is evident and stunning… :ph34r:
And i was reported about this before - mix of separated tracks in external mixing app sounds different (better) than mix of same trax in Renoise.

Hope we’ll find the root of this problem by common efforts !

Except that frimium loaded different presets in 1.20 than in 1.5, which is a little bit scary, I could not see that any frequencies are lost in 1.5 or 1.2. This was after pressing “zero” on every single frimium instance to reset the preset.

If you do this test at home, make sure that all presets are flat. Frimium is not only a analyzer, but also a filter plug.

As I wrote before:
Lets stay polite. This was not a mixing test, but a test of the whole complex environment, including every single dsp plug that Renoise has. I bet Renoise internal filters are the problem here, because we do have done a lot of mistakes when it comes to backwards compatibility.

If we want to find out if the mixing is the cause, we have to test a song that does not use any dsps at all.


Plugin is named “Firium”, try to be attentive. And it is bag report rather then provocation. Please do not try tangle me, because i know renoise have this bug anyway. Becouse i work with it very many time. And high frequences in latest versions is worsed in any events… Or can be only one opposite situation - older versions have bug which perfects sound. But this suggestion sounds funny…

And of course it is the filter plugin. I make complex and hard test with very many filters to do testing external dsp and internal mix of renoise.
Sting you can’t understand me right.

This was no offence. Lets stay on topic and try to solve this problem.

Has anyone else tested EternalEngine’s test songs? Any help is welcome.

“This was no offence.”

What is “no offence”? :huh: Firium works normal. It is normal 50 band Pro Equalizer. Differences in high frequaences after hard dsp and mixing is evident. Sample do not have internal filters and it is white_noise sample. You can transpose it up or down to halftone and nothing will be happened. Becouse it is white noise and frequency calculation is does not matter in this situation. If you want you can change this sample to any VSTi white noise generator. And nothing will don’t changed in results of the test. Old versions of renoise and primary version 1.20 have much more high frequencies in hard dsp calculations.

“Lets stay on topic and try to solve this problem.”

What you want more? I show you all what i can show about this bug.

I did test the songs and found no difference in frequency analysis.

This is live recordings from renoise 1.20, 1.25 and 1.5. Using a E-mu 1820m card at 48khz, and dx 48khz in renoise.

I could not load test song 1 into renoise 1.2
Test song 1:

Test song 2:

In other words there was no audible or visual difference in the renoise versions.

I also testet this using a Audigy 2 card. Same result.
I can try doing a render test tomorrow.

Can someone else try this too?

Could you post a freq. analysis too?


What does the green curve refer to? Looks quite distorted.

We both really have a communication problem.

“No offence” means that I did /not/ wanted to say in my previous post that I am right, and you are wrong. I just wanted to note that the plugs in the testsong had different preset settings when loading them in 1.5 and 1.2, and one should take care about this when doing the tests.

Ok… It does not matter. I did not scan a range for this analysis. It was a instant snapshot. All versions vary like that. So they are equal.
Was just too lazy pressing the record button and then scan a range. Then you will get a much smoother curve.
Something to do tomorrow. I’m of to bed. ;)

First. I can’t prove this, but i think your results is mistake or falsification. Becouse i can heard difference in results even with audigy2 and emu is better sound card. All know about that.

Second. Why you do external “live recordings” if you can do internal analise of wav file? I specially for you and for all send link to “tapeit” vst plugin. It can record (render) internal sound of renoise demo versions, from master channel to wav file. Without changes of digital analogue transformations.

Ok, i can show to you real results of this test…

Test song 1 maded with Renoise 1.25 and show differnce of Renoise 1.25 and Renoise 1.5. It is dosn’t matter in this time. And i show to you analise of test song 2 which rendered with tapeit plugin from master channel to 32 bit wav file.

Renoise 1.20 test song 2

Renoise 1.5 test song 2

As i write before and as you can see different is evident.

You think Firium work different in 1.2&1.5 versions? Why? You can motivate this? I check it and i think it work normal in both versions.
If i do test with other EQ you will again say that they work on differring presets?

I only tested the Test song 2 because test song 1 has instrument filter enabled (reso):

Test Song 2 rendered with tapeit 32bit 48khz:

Its scanned about 225ms.
Green = 1.5
Red = 1.25
Blue = 1.20

As you can see there are no frequency loss in my tests :ph34r:

If this turn out to be a bug that only appear occasionally, then I guess we need more ppl to test this.
For me everything is ok.


Pisj, sorry but your test’s looks unbelievably. I can believe what different versions can do differnt CPU&Memory load, it is normal situation.
But different render on many computers it’s somthing from science fiction. :P

Ok, let’s wait when anyone other test this problem.

P.S. Taktik I understand it maybe not in your interests, but maybe you can publish spectrum analise graphs from your work computer?

No there are no frequency losses here, but it is obviously clear that 1.5 seems to vary quite a lot of DB under the other two (1.2 and 1.25) when it involves very low (<~15700 Hz) or very high (>~22600Hz) frequencies.
EternalEngine’s results chop even down lower.

My sorry for all, i use don’t last demo version of Firium plugin. Today i download last version
and it have near to equal results in 1.2, 1.25 and 1.5 versions. But results anyway have some little difference, in the end of graphic and in global volume level.

I capture big full windows from Adobe Audition 1.5 with maximum fft size 65536 and reference -20db, lenth of tested sound 300ms:

Renoise 1.20 test song 2:

Renoise 1.25 test song 2

Renoise 1.5 RC2 test song 2:

PS. First time i made 16 bit tests. Pleese wait i change it to 32 bit…

2 EternalEngine:
So, as we can see - difference is minimal.
But what about those your audio examples, where diffrence was obliviously audible ? Why it happens ? And how can it be at all ?

Edit: Now it’s much more visible with 32 bit analysis graphs

On this graph we can see the same frequency bump at 13-14khz which are absent in 1.5 !
It’s very strange… Maybe it was a bug of older versions ?
Is it possible this little freq. peak affects so audible differense when summing all channels and fx together at mixing ?