Convolver applies dc offset filtering to impulse

Help, Support & Bugs
1

The convolver processes small impulse files ( cabinet etc…) almost identical to most other imp.reverbs .

However sometimes when loading certain imp.files , there is still a big difference …especially medium sized rooms etc…

It’s mostly the first portion of the reverberated output that makes the difference , ( obvious when rendering a 1 sample impulse trough convolver and comparing the rendered file )

When other reverbs sounds distant, behind you; convolver sometimes sounds nearby/centered , ( so obvious when using headphones)

I noticed that convolver is applying dc offset filtering ,which might clarify the different results compared to other reverbs .

As a test I created an impulse file of the psp pianoverb vst effect .( process the dsp effect with a 1 sample impulse )

The rendered file ( see screenshot ) has no negative energy .

When this file is loaded in the convolver , the convolver is applying dc offset filtering .

Took it a step further and loaded a sample with only +1 energy , a pure dc sample .

As expected ,convolver applies high pass filtering.

I don’t know whether the filtering is done to the input signal or the imp.file , or both.

The effect is most noticeable when processing a oneshot decaying rampwave ,positive only , convolver loaded with the provided imp.file .

I can totally understand the benefits of this , minimizing the risk of dc offset .

On the other hand , this might also modify the processed output .

Maybe a option to turn off dc filtering …?

Provided is the impulse file and an other freeware imp.reverb t

Don’t get me wrong , the convolver has definately improved since it’s first appearance , but it still has some difficulties

with the beginning of certain imp.files …

Link to an amazing freeware imp.reverb.( cancer research donation ware )

http://www.liquidsonics.com/software_reverberate_le.htm

renoise file

https://app.box.com/s/dlebgekynf0tneuvbspbcpllyfpykrhn

and the impulse file , (positive energy only )

https://app.box.com/s/tt089pu71dele1qx6t0xw1b851qatnhb

2

I haven’t ran into this personally, because none of the impulses I use are like that. Have to ask, what is the source of the piano impulse? Is it literally just an impulse of the inside of a piano, and does it sound good on piano samples? Gotta try it when I get home from work.

3

It’s an impulse made of a vst effect called ’ pianoverb ’ by psp ( freeware )

You can make impulses of vst effects by drawing a 1unit sample ( make the file long enough … seconds ) and then click apply fx .

Back on topic ,

bump

4

might as well bump this

5

what’s “cool” about the renoise dc filters is that they shift phase so hard that they become an interesting sound design tool. All 4 are like this, sample editor, dc device, distortion, convolver…they generate fancy zipper noise sounds a bit like rubberducks if you distort the crap out of the noise floor with multiple distortion devices. That is why I have already known, that there is the DC filter in the convolver device and shift distortion pkchhhh-p.ziiouubwgn

Is the DC filter applied to the input, or to the IR sample? I’m too tired to try to think of what difference it would make (maybe none?).

Don’t drive certain things in renoise too hard with asym input, some devices don’t like this and probably the DC filters are in there to mitigate the risk of overflow.

I would also really love being able to disable those filters at own risk.

6

I still think maybe the difference in sound/perception could be resampling, not so much the DC filter. I just rendered a beat using some convolvers 2 times, 44.1 and 192, and converted the 192 to 44.1 as reference. The 44.1 render sounds a bit more dull, less spacious somehow. The 192 and interestingly also the conversion to 44.1 had some extra depth to them, noticeable on a/bing the sound. Like in the hi res render was some faint layer behind the surface, that is more or less gone or much weaker in the low res version. With open back headphones. The IR used was a custom crafted IR with uncorrelated stereo white noise, such is not extra steady on the attack but can sound artificially wide/deep. Difference was very subtle though, and seemed only in the high frequencies (snare tails). It could ofc also be distortions in the processing chain changing sound because of the high sampling rate.

So maybe, to tune the convolution reverb, it might be a good idea to work/render on higher rates like 96khz? Ofc this will put quite some extra strain on the CPU depending on your project.

Or is the difference you experience more drastic, like in comparison to other convolution plugins…have not tried abing such yet…and I thought the difference is really rather subtle, on an average stereo you probably won’t notice, but on real hi q rigs or studio monitors or the like…