Is there a “Band Shelf” filter? It seems like it would follow naturally from the others. It’s possible to use the EQ to arrive at something similar, but it’s harder to set up and automate.
Hm band shelf is different to band pass, right? The slope of a band shelf is very straight, the band pass is baggy to the top… I guess you can simulate a band shelf then using the 8th order filters of renoise in bandpass mode? Anyway, something currently is moving secretly in the dark, I guess in the next days we will have Renoise 3.1, and it has some much more filter options.
there are no shelf filters in the new filters anymore, so you can only use the old ones
Oh, yeah I thought band shelf is almost the same as bandpass, only flat. That seems to be wrong.
MattD: Maybe you can use the deprecated “old” but still good filter device in Redux, but copy paste this to the dsp bar:
[SPOILER]
<?xml version="1.0" encoding="UTF-8"?>
<FilterDeviceClipboard doc_version="0">
<DeviceSlot type="Filter3Device">
<IsMaximized>true</IsMaximized>
<IsSelected>true</IsSelected>
<SelectedPresetName>Init</SelectedPresetName>
<SelectedPresetIsModified>true</SelectedPresetIsModified>
<IsActive>
<Value>1.0</Value>
<Visualization>Device only</Visualization>
</IsActive>
<Type>
<Value>3</Value>
<Visualization>Device only</Visualization>
</Type>
<Frequency>
<Value>0.280000567</Value>
<Visualization>Device only</Visualization>
</Frequency>
<Q>
<Value>5.45999908</Value>
<Visualization>Device only</Visualization>
</Q>
<Gain>
<Value>-60</Value>
<Visualization>Device only</Visualization>
</Gain>
<Inertia>
<Value>0.0078125</Value>
<Visualization>Device only</Visualization>
</Inertia>
<Model>Butterworth 4n</Model>
</DeviceSlot>
</FilterDeviceClipboard>
[/SPOILER]
Works here. Btw. what for do you need such band shelf? Some kind of manual phaser?
EDIT:
Err, no I was right, Gova was wrong!!!
BS means “bandstop”, simply inverted bandpass. A band shelf is really some kind of bandpass, only with a “flat curve”. So maybe use bandpass butter worth 8p:
[SPOILER]
<?xml version="1.0" encoding="UTF-8"?>
<FilterDeviceClipboard doc_version="0">
<DeviceSlot type="Filter3Device">
<IsMaximized>true</IsMaximized>
<IsSelected>true</IsSelected>
<SelectedPresetName>Init</SelectedPresetName>
<SelectedPresetIsModified>true</SelectedPresetIsModified>
<IsActive>
<Value>1.0</Value>
<Visualization>Device only</Visualization>
</IsActive>
<Type>
<Value>2</Value>
<Visualization>Device only</Visualization>
</Type>
<Frequency>
<Value>0.450000554</Value>
<Visualization>Device only</Visualization>
</Frequency>
<Q>
<Value>11.9000006</Value>
<Visualization>Device only</Visualization>
</Q>
<Gain>
<Value>-60</Value>
<Visualization>Device only</Visualization>
</Gain>
<Inertia>
<Value>0.0078125</Value>
<Visualization>Device only</Visualization>
</Inertia>
<Model>Butterworth 8n</Model>
</DeviceSlot>
</FilterDeviceClipboard>
[/SPOILER]
not sure about that, jurek,
a shelf implies a shelf, not -inf gain outside the rangethat a bandpass would (imperfections aside)
Uh I only know bandshelf filters, which are rarely offered in such, as part of some graphical equalizers. Isn’t it like some defined “plateu” in spectrum that will be boosted or dipped, in contrast to the usual peak/notch filters that will have a tip and bell shape instead of a flat range?
Trying to build such yourself with other filters could mean having the uneffected signal mixed with a bandpassed signal, so a range gets boosted. So: done in parallel, i.e. with send channels. Or mix uneffected with notch filtered, and compensate overall gain, to have some dip range in the spectrum. This sounds too easy though, I think the final result won’t be as cool as a real filter built for this, because filters usually effect phase so you won’t have the desired boost/dip response, but some strange parts hard to predict when mixing parallel signals because of cancellations.
This is the approximation with EQ. What I have in mind are situations where a track that occupies a narrow band can’t be heard through a track that occupies a wide band. If this were a filter I could imagine automating it with key tracking and a signal follower.
This is as far as redux automatable shelf filters can get you, if no realtime automation is ok, there is the convolver…
I’ve made filter chains that would filter out specific harmonics via a keytracker, but there’s the problem of phase it endup messing up a lot of things
Is this the time we ask the renoise team for a linear phase zero delay filter? ha, i’d love that
This is the approximation with EQ. What I have in mind are situations where a track that occupies a narrow band can’t be heard through a track that occupies a wide band. If this were a filter I could imagine automating it with key tracking and a signal follower.
EQ is one approach. Instead of the EQ bells just use the shelfs of the EQ. Set the low shelf to the top frequency and the high shelf to the lower frequency of your band range. Connecting the gain of both to a Hydra lets you adjust your band shelf with a single parameter then.
Second way is using 2 shelf filters (of what type you ever prefer) in the same way: low shelf with top frequency, high shelf with lower frequency. Usage of Hydra ofc makes things easier here too.
So in fact a lot of quite different band shelfs are possible and you could easily setup shelf combinations far beyond a common band shelf.
Edit: And of course add a gainer synced via hydra to eleminate volume changes. 
Edit 2: This is a quickly setup Doofer for the band shelf via EQ. download Didn’t test it, but it should work this way. Grab it, as long as its up. I won’t guarantee it’s up forever and might be gone with the next webspace clean-up.
The setup with seperate filters would be a bit more complex, since it needs a filter switch for positive/negative gain + a few more gainers to compensate the gain changes, but in general works the same way. To be honest, I wasn’t up to work this out now. 
But shouldn’t be too hard with some LFO to switch filter types.
Is this the time we ask the renoise team for a linear phase zero delay filter? ha, i’d love that
Linear phase zero delay? Keep on dreaming. Filters are either low delay but phase shifted, or long delay but linear phase. Phase vs. delay is a tradeoff point.
I’ve tried mixing a notch filtered with dry, and then gaining both in x-fade style. Kind of works for shelving out a band. I’ve only tried with white noise. The xfade volumes work when using the send gains, but not really linear style, might be correctable with curves. Gain seemed very consistant. I’ve noticed phasing out in the spectrum, though, dips in the spectrum around the corner frequencies of the notch filter. Butter n8 filter in renoise 3. Setup needs a send channel, sending the track output there once, filtering, and then sending again to the same send cannel, having the 2 send-device gains controlled inverse proportional to set the amount of reduction in the band.
The opposite, boosting a range by combining with bandpassed signal, sounded no to well, very phaser-alike, much more than dipping via crossfaded notch.
Normally for your purpose, people would use some cool plugin that does banded compression. It’s not really some “musical” effect, but rather something you do when mastering, to chisel out clarity of the end mix.
OopsIFly: But convolution can be zero delay and you can convolve with an IR that is linear phase, super tiny delay linear phase (wait maybe it is truly 0 delay… gah I don’t want to look at fft math, you win this argument :P)