I presented this question to Computer Music magazine and they answered it, but I missed the issue (thanks to my local Borders closure). Months went by. I still want to know. I've seen people say "the filters on [n] are awesome" and "[n]'s filters are poor," and "the [n] filter is way more musical than 's filters." Et cetera.
So, having played with many synths and swept their filters and whatnot, I'm not really getting the difference between one filter and another. I get the filter type is different but what makes one company's or product's [n]-pole filter better or "more musical" than another implementation? Avoiding zipper/stepping? But then what about analog where there's no stepping by analog nature?
Help me understand and hear it?
What makes one filter better than another or "more musical?"
-
- KVRAF
- 6113 posts since 7 Jan, 2005 from Corporate States of America
- dysamoria.com
my music @ SoundCloud
my music @ SoundCloud
-
- KVRAF
- Topic Starter
- 6113 posts since 7 Jan, 2005 from Corporate States of America
Well, that's bizarre. I thought there would surely be lots of opinion on this at the very least!
- dysamoria.com
my music @ SoundCloud
my music @ SoundCloud
-
- KVRer
- 21 posts since 1 May, 2012
I don't think this qualifies as "musical" but some filters provide very little (or no) phase shifting or delay, which can be really important if you're doing things like parallel compression.
(Wikipedia has a good article, but I can't link to it because I don't have enough posts)
In general, maybe it just has to do with a combination of non-musical things, like whether the parameters are smoothly interpolation, have optional hysteresis to make fast filter sweeps sound nicer, etc.
Just throwing this stuff out there as a starting point. I think you'll find a lot of music technology stuff is very subjective. It doesn't mean there's no truth in it, it's just that "sounds better" or "sounds more musical" can be a tough thing to quantify, eg. a combination of tiny little things.
(Wikipedia has a good article, but I can't link to it because I don't have enough posts)
In general, maybe it just has to do with a combination of non-musical things, like whether the parameters are smoothly interpolation, have optional hysteresis to make fast filter sweeps sound nicer, etc.
Just throwing this stuff out there as a starting point. I think you'll find a lot of music technology stuff is very subjective. It doesn't mean there's no truth in it, it's just that "sounds better" or "sounds more musical" can be a tough thing to quantify, eg. a combination of tiny little things.
Oscillicious Audio Labs | Slice Beats with BeatCleaver.
-
- KVRAF
- 5223 posts since 20 Jul, 2010
For a synth to be musical it needs to work with the oscillator in a way that brings out a wide range of different subtleties at all extremes of parameter ranges. When there's no resonance, a sweep should feel satisfying and interesting (subjective) so often phase issues and other colouring limitations can work in a filter's favour. Some filters just sound like almost a volume control with no resonance, this is not ideal IMO.
When resonance is applied, it should sound like a part of the sound, and not a seperate entity. If distortion happens it should be controllable over a wide range and not have a "lumpy" range - i.e. as you turn the knob, the sound changes at a constant rate. The composition and harmonics of the distortion are also important, I think even harmonics are regarded as more "musical" but don't quote me on that.
Really, the whole thing is subjective. Some filters just cut and boost frequencies. Others speak to me. Try comparing the filter in something like Oatmeal with the filter in Diva, Saurus, Charlatan, etc and you should get a rough idea. But then, your preference may simply be different than mine.
When it comes to EQ shapes, broad curves are considered more musical than steep ones. I also happen to generally prefer shallow cutoff filters, for example, resonant 6db/oct filters which are more colourful than 24db/oct ones - though this depends entirely on the implimentation and specifics... For example you want a tight steep filter for a punchy bass.
When resonance is applied, it should sound like a part of the sound, and not a seperate entity. If distortion happens it should be controllable over a wide range and not have a "lumpy" range - i.e. as you turn the knob, the sound changes at a constant rate. The composition and harmonics of the distortion are also important, I think even harmonics are regarded as more "musical" but don't quote me on that.
Really, the whole thing is subjective. Some filters just cut and boost frequencies. Others speak to me. Try comparing the filter in something like Oatmeal with the filter in Diva, Saurus, Charlatan, etc and you should get a rough idea. But then, your preference may simply be different than mine.
When it comes to EQ shapes, broad curves are considered more musical than steep ones. I also happen to generally prefer shallow cutoff filters, for example, resonant 6db/oct filters which are more colourful than 24db/oct ones - though this depends entirely on the implimentation and specifics... For example you want a tight steep filter for a punchy bass.
http://sendy.bandcamp.com/releases < My new album at Bandcamp! Now pay what you like!
-
- KVRAF
- Topic Starter
- 6113 posts since 7 Jan, 2005 from Corporate States of America
Is this all about sweeping a filter, then? I've almost never used that in my music.
- dysamoria.com
my music @ SoundCloud
my music @ SoundCloud
-
- KVRAF
- 8686 posts since 24 May, 2002 from Tutukaka, New Zealand
For me, musicality in a filter is most of what Sendy wrote. Wuth the addition that I like a bit of movement in a filter - i.e. not necessarily strictly linear and predictable (for example with some of the old analogue synths, if you play the exact same patch and same note, the filter will never sound exactly the same each time, even though it will obviously be very very similar).
I'll expand on the bit about it sounding part of the sound too. There are many filters that when you push them into self resonance, just give a whistly sine wave over the top, that bears no relation to the oscillator actually being played. You can play a saw wave, a sine wave, a pulse or even some mega hoover, and it'll still just give you that bland piercing whistle over the top. Mostly digital filters do it, but analogue ones aren't immune either. E.g. Juno106 filters do exactly that - the synth has its uses, but pushing the filter is not one of them. If you have it in self res and play almost any octave, the riff will generally be ear piercing and irritating, whereas if you play something like an OSCar (both the analogue and the VSTi) you can push the filter res up to max and play a riff with most filter types without it being annoying or shrill. It'll still make the riff sound musical, i.e. gives it movement or character and interest.
It's especially something you notice when you push a filter on a patch and play in the high octaves. The filter either becomes even more whiney or just disappears completely. Play most VSTi in octave 4 or 5 with a heavy filter res and you almost forget that a filter is doing anything. Same with Junos etc. But play an OSCar (or impOSCar) in octave 5 on a heavy highpass filter setting and it will still swirl around without making you want to scratch your earhole.
I've noticed over the years that I like 24db/Oct filters less - they tend to be less musical. Maybe it's just the standard implementation of them that's been copied from a bad example. Whereas synths like the Yamaha CS range, Korg MS range used 12 db/Oct and they are much more musical (partly because of a nice warm distortion in the filters or general sound architecture too). Then you have things like the 303s - 18dB/Oct and very characterful. I used to wonder if it was just that 24dB/Oct emphasised frequencies too much, but some 24dB/Oct filters can still work musically. Roland 101/202s can still sound musical at highish res settings. I don't know about filters enough, but maybe some have a wider Eq curve on the filter? Certainly that works with straight Eqs - shorten the Q down narrow and boost it, and it gives you a nasty piercing noise. Whereas a wider Q allows much higher boosting.
I'll expand on the bit about it sounding part of the sound too. There are many filters that when you push them into self resonance, just give a whistly sine wave over the top, that bears no relation to the oscillator actually being played. You can play a saw wave, a sine wave, a pulse or even some mega hoover, and it'll still just give you that bland piercing whistle over the top. Mostly digital filters do it, but analogue ones aren't immune either. E.g. Juno106 filters do exactly that - the synth has its uses, but pushing the filter is not one of them. If you have it in self res and play almost any octave, the riff will generally be ear piercing and irritating, whereas if you play something like an OSCar (both the analogue and the VSTi) you can push the filter res up to max and play a riff with most filter types without it being annoying or shrill. It'll still make the riff sound musical, i.e. gives it movement or character and interest.
It's especially something you notice when you push a filter on a patch and play in the high octaves. The filter either becomes even more whiney or just disappears completely. Play most VSTi in octave 4 or 5 with a heavy filter res and you almost forget that a filter is doing anything. Same with Junos etc. But play an OSCar (or impOSCar) in octave 5 on a heavy highpass filter setting and it will still swirl around without making you want to scratch your earhole.
I've noticed over the years that I like 24db/Oct filters less - they tend to be less musical. Maybe it's just the standard implementation of them that's been copied from a bad example. Whereas synths like the Yamaha CS range, Korg MS range used 12 db/Oct and they are much more musical (partly because of a nice warm distortion in the filters or general sound architecture too). Then you have things like the 303s - 18dB/Oct and very characterful. I used to wonder if it was just that 24dB/Oct emphasised frequencies too much, but some 24dB/Oct filters can still work musically. Roland 101/202s can still sound musical at highish res settings. I don't know about filters enough, but maybe some have a wider Eq curve on the filter? Certainly that works with straight Eqs - shorten the Q down narrow and boost it, and it gives you a nasty piercing noise. Whereas a wider Q allows much higher boosting.
-
- Banned
- 12367 posts since 30 Apr, 2002 from i might peeramid
i've got a feeling it could be about emphasising the desirability of some products in the marketplace... do they still say things are phat in print?
you come and go, you come and go. amitabha neither a follower nor a leader be tagore "where roads are made i lose my way" where there is certainty, consideration is absent.