Does anyone know how to use all pass filters to restore the phase shift from filters.

2»

Comments

  • Studiowaves
    Studiowaves Member Posts: 734 Pro
    https://community.native-instruments.com/discussion/comment/219943#Comment_219943

    I concur, it's not possible. I think this pretty much settles my problem. All in all these 3 filters should work fine on the 6 possible outputs as lone as each output uses the same filter. That way there will be no interactions due to phase shifts. I guess in reality we strive for simple designs with the lowest CPU to get close to what we are after. After a bit of deliberation have a choice of one of these 3 filters along with a non filtered output should cover things pretty well. I custom made 3 pseudo shelving filters out of three LPF's. The shelving are doesn't go perfectly flat but they don't need to. I suspect they use less cpu like this compares to true shelving filter too but I haven't tested it.

    20-20k sweep with filter using one, two and 4 pole LPS's with pseudo shelving.ens
  • Studiowaves
    Studiowaves Member Posts: 734 Pro

    Well, the ensemble is there to make analysis easier. Maybe you can come up with something. I'd like to see the Gauss filters. What I'd really like though is the Waldorf filters that Steinberg's Halion sampler use to have. They actually sounded pretty realistic on some instruments. Not sure how they work and I've had no luck trying to emulate them with Reaktor. Maybe you know how to do it. I'm quite sure many of us would like to have them in the filter library. I had a chance to work with them at studio about 20 years ago. Used the Waldorf's to filter Flugal horns and it really sounded good. I remember how popular they were at the time.

  • Tr97
    Tr97 Member Posts: 26 Helper
    linear phase single crossover 2.ens

    Here is a basic ensemble with a crossover made of this Gaussian Lowpass in some variants (LP made of 4 moving averages (has ripple in the stopband), LP made of 8, variant with steeper highpass and variant with even steeper highpass (think these two are not so good, cause they resonate)

    Some disadvantages: They are not normed at the moment (so the cutoff frequency is not corrected and varies between types), there are very coarse cutoff frequency steps on high cutoff frequencies (this is because of the moving averages taking only integer sample lengths, for a more professional thing maybe the formulas how to distribute the delay times to the instances could be optimized (ran out of ideas) or even switch to direct convolution (on <20 samples this should be ok)), high latency, maybe a bit weird behaviour on start of on changing frequencies and some more.

    I've seen some screenshots of the Waves L3-16 limiter, the rolloff curves seem to be very similar to this, I am not sure, but it would make sense.

Back To Top

Categories