A Quick Analysis Of Renoise Filter3 Modes

[dblue]

If you find the various modes in Renoise’s filter device to be a bit confusing, and you’re wondering how to get a standard filter slope of 12dB, 24dB, 48dB, etc., then hopefully you will find this post useful.

Using the excellent RubberFilter by Christian Budde and Voxengo SPAN to analyse the frequency response of the various filter modes, we can see that:

• 12dB = Renoise Filter 24dB 4Pole ( Not a perfect match, but pretty damn close )
• 24dB = Renoise Filter Butterworth 4n
• 48dB = Renoise Filter Butterworth 8n

(In all cases filter resonance was set to zero)

You’ll see in the screenshots below that RubberFilter’s 24dB mode is a perfect match for Renoise’s Butterworth 4n mode, and that RubberFilter’s 48dB mode is a perfect match for Renoise’s Butterworth 8n mode. This comes as no big surprise since RubberFilter itself is also built on Butterworth filters.

Renoise 24dB 4Pole Highpass 1000Hz vs RubberFilter 12dB Highpass 1000Hz

Renoise Butterworth 4n Highpass 1000Hz vs RubberFilter 24dB Highpass 1000Hz

Renoise Butterworth 8n Highpass 1000Hz vs RubberFilter 24dB Highpass 1000Hz

Unfortunately, achieving a matching 6dB slope was not so simple. At first I thought Renoise’s 24dB Moog might be a good match for a 6dB slope, but here we can see that this is not the case:

Renoise 24dB Moog Highpass 1000Hz vs RubberFilter 6dB Highpass 1000Hz

I suspect that Renoise’s 24dB Moog might actually have a response that is closer to a 3dB slope, but unfortunately RubberFilter does not have a 3dB option that I can test against, and at the moment I do not have any other suitable filter plugins that I trust.

Either way, it is actually possible to get an almost perfectly matching 6dB slope from Renoise’s filter, but it does require a little bit of experimentation to adjust the frequency until it produces the same result. In this case I had to increase Renoise’s filter frequency from 1000Hz to 2500Hz:

Renoise 24dB Moog Highpass 2500Hz vs RubberFilter 6dB Highpass 1000Hz

Anyway… I hope this is helpful to some people. The moral of the story is: don’t be afraid to experiment. Also trust your ears more than your eyes… if it sounds ok, then it’s probably ok!

[kazakore]

Good post. Doesn’t take into account resonances, which behave very different for the three types of filter, but gives a good idea when using them for shelving high/low frequencies.

@dblue

I suspect that Renoise’s 24dB Moog might actually have a response that is closer to a 3dB slope, but unfortunately RubberFilter does not have a 3dB option that I can test against, and at the moment I do not have any other suitable filter plugins that I trust.

As a single pole filter is 6dB/Oct it is unlikely many people will model a filter that is not based on what exists in the real world, although there probably are exceptions. Multi-pole filters are multiples of 6dB/Oct.

[dblue]

@kazakore

As a single pole filter is 6dB/Oct it is unlikely many people will model a filter that is not based on what exists in the real world, although there probably are exceptions. Multi-pole filters are multiples of 6dB/Oct.

Yeah, I figured this would be the case. I seem to remember seeing a 3dB slope mentioned somewhere at some time, but it may have just been in a paper that someone had posted online, or as a comment in a bit of source code long ago, who the heck knows. I know a little bit about this stuff, but I will openly admit that I am definitely not a DSP guru when it comes to the strange world of poles, unit circles, imaginary numbers, etc. Still got a lot to learn

[kazakore]

Oh I’m fairly sure it would be a lot easier to make a digital filter with any slope you so desire, but it may be an impossibility to make it out of real world electronic components, and thus modelling it to a real design becomes even harder (I would imagine.)

Personally my understanding of electronics is fairly above mine of DSP coding so we probably come from slightly different sides though.

[kasmo]

thanks for this dblue, great work

[Foo (mr_mark_dollin)]

Fascinating post dblue, thanks.

[s-n-s]

nice post for sure

hey Dblue have you seen that theres a beta kind of thing out for span 2:)?

[dblue]

@s-n-s

have you seen that theres a beta kind of thing out for span 2:)?

Yup, I have it here, I just haven’t made the switch to it yet.

[It-Alien]

dblue, you should make some filter presets and send them to taktik so he can add them to the filter factory presets

[weedwalker]

good post.

but i dont get why we call it 24dB 4Pole if it’s actually a 12dB filter.

but anyway, good to know we have a 12db filter.

[dblue]

@It-Alien

dblue, you should make some filter presets and send them to taktik so he can add them to the filter factory presets

Perhaps all the graphs and stuff made this seem more complex than it really is. Literally all I was saying is that the default 24dB 4Pole filter with resonance set to zero will give a 12dB/Oct slope, and the Butterworth 4n with resonance set to zero will give a 24dB/Oct slope, etc. There’s nothing more interesting than that really going on here. Mainly, this was just intended to be a response to some posts and feature requests that sometimes pop up, where people ask for features which already exist.

Examples:

More High-Pass Filters In Renoise

High-pass filtering is probably the most used processing for recorded parts (to remove LF rumble). In Renoise it’s a bit awkward, because there are no filters in the native EQs (low-shelf isn’t quite the same) and the filter is 24db/octave and high resonance at minimum, which is very aggressive and mostly good for shaping synth sounds. Getting 6db/oct and 12/oct filters without external GUIs would be really useful for me. Any chance to get these implemented? I’m not sure but didn’t the old fil…

http://www.renoise.com/board/index.php?showtopic=20501

The confusion seems to stem from the fact that the filter names in Renoise are not labelled in a way that people might expect. Maybe people are following tutorials they’ve seen online or in magazines, which mention the need to use a “12dB/Oct highpass” or “24dB/Oct lowpass” to achieve a certain type of sound, and then when they come to Renoise they do not immediately find such an option. In these cases, I can understand some of the confusion and potential frustration that people might be experiencing.

For me personally, it only takes a few moments to experiment and see what is actually happening to the sound, but maybe it’s a good idea to think about renaming the filter modes slightly to avoid this confusion in the future?

[vvoois (vV)]

@dblue

For me personally, it only takes a few moments to experiment and see what is actually happening to the sound, but maybe it’s a good idea to think about renaming the filter modes slightly to avoid this confusion in the future?

It depends whether naming conventions are patented or trademarked…
Like the Cabinet simulator is definately a good example. It contains different names for known algorithms pure to avoid legal issues.
I’m not saying it is the same for filters, but they also rely on algorithms and perhaps they act different because of algorithm patents perhaps (perhaps that’s why they are not 100% exact copies)?

[Akiz]

We should remind it to Renoise 3

[SynthisterNation]

I feel like I’m missing something important here with the photo’s being down.

[dblue]

I feel like I’m missing something important here with the photo’s being down.

Indeed. Imageshack seems to have deemed those images trash-worthy — perhaps because this thread hadn’t been viewed regularly enough, who knows. I no longer have the images stored on my drive(s), either.

Ah well, it’s perhaps not the best idea to get so lost in the technical details anyway. We’re trying to make music here, not perform science experiments (at least not all the time!)