MichaelBenjamin wrote:martinvicanek wrote:Your filters are unoptimized direct form 1 - that's quite wasteful. You could improve CPU load by an order of magnitude with optimized TDF2 filters, say.
wlangfor has a certain idea about use of certain math and opcodes for the signalpath, so i guess that could be intentional for a better/different sound regardless of cpu? Not everything in digital music has to be cheap in cpu. Sometimes the cycles need to burn like new eve fires.
Hmm, yes to be honest I like the unique sound of these filters. It really did not come down to optimization. However, I still want to attempt serialization. That is to change the five co-efficients to a linear min/max change based upon sample rate. My rationale is that I love the sound of these filters.
I like the TDF2, they're good but the sound is also quite unique. To be honest I find they have a very professional sound that is also a bit dry like as it is solid state. So, in regards to the resonance / q ratio being too wide it is due to My extensive testing of this EQ mixing vocals and drums, classical violins and plucks.
Though One could argue that it is less technical I felt this was a happy medium between some technical prowess and usability. Thanks Martin, and it must have been annoying seeing a parallel filter being made extremely slowly.
martinvicanek wrote:MichaelBenjamin wrote:wlangfor has a certain idea about use of certain math and opcodes for the signalpath, so i guess that could be intentional for a better/different sound regardless of cpu?
Not sure, the DF1 is the first thing you would grab without giving it too much thought. There are certainly cheaper and better filter structures in terms of quantization noise etc.
MichaelBenjamin wrote:Not everything in digital music has to be cheap in cpu. Sometimes the cycles need to burn like new eve fires.
I agree! Sometimes we get carried away with optimizing stuff. It should not be the main design aspect.
Well, ultimately I was going to add some special two channel crossfaders so that if the band is at zero and not being used I can ensure that the co-efficient is turned off.
Look in the My beta EQ's post; It's within the shelf EQ. It allows You to switch between two signals. The Boolean would be driven by !==0.
I'll optimize a bit more, I was pressed for time. Thanks to both of You for Your input and I can see the reason for scrutiny too. All in all, I hate to blame commercial EQ's; If I'd had LA Bands in the past I wouldn't have sounded so Noob like some of My earlier tracks sound.
That's why there is some wide Q constraints so that there is less chance of distortion and clipping. It came down to extensive testing.
EDIT: And yeah, I will add some presets with a manager.