Building Filters?

Author
Wizky
Max Output Level: -90 dBFS
  • Total Posts : 11
  • Joined: 2005/07/07 16:33:57
  • Status: offline
2006/10/29 09:47:17 (permalink)

Building Filters?

Is there somehow any possibility to "reverse engineer" a filter and then implement it in Rapture's sfz format?

E.g by feeding the synth with a sine wave through the filter (in various settings) and using a spectrum analyzer or some other tool to analyze the frequencies? Or do you need to have the specs of the particular filter to get as close as possible?
I'm aware of that it would be impossible to get an exact replica but it would be great fun trying to get the most out of it.

Just curious since I'm very fond of my K-Station filter. It's a 12/24 pole filter, but it has certain characteristics....
#1

5 Replies Related Threads

    b rock
    Max Output Level: 0 dBFS
    • Total Posts : 8717
    • Joined: 2003/12/07 20:31:48
    • Location: Anytown (South of Miami), U.S.A.
    • Status: offline
    RE: Building Filters? 2006/11/01 15:59:47 (permalink)
    Is there somehow any possibility to "reverse engineer" a filter and then implement it in Rapture's sfz format?
    Interesting. Very interesting indeed. In the past, I'd been experimenting with existing filter and EQ stacks to emulate certain characteristics. But why not convolution? SF and Audition, or maybe the free SIR download will give you the basic tools. The Dimension Pro XP-1 pack pointed the way to getting it into .sfz, with it's Cabinets and Amp Simulations.

    If I'm not mistaken, the K-Station allows you to run the audio input through the filters. That way, you could construct an impulse externally and feed it through. If not, you may have to build a noise burst or a sweep of sine .wavs with the oscillators as the impulse. Once you've extracted that and gotten your impulse response, save it as a .wav, and point an .sfz (with the tdfir <effect> in it) at the results.

    Changing that emulated filter over time would be the trick. I suppose you could "animate" the filter in the K-Station and capture that, but I'm thinking that the result will be a static snapshot in .sfz. Thanks for the *spark*. It's a sketchy theory, but I'm going to see what shakes out.
    #2
    René
    Max Output Level: -68 dBFS
    • Total Posts : 1103
    • Joined: 2004/01/06 13:15:57
    • Status: offline
    RE: Building Filters? 2006/11/01 17:57:23 (permalink)
    You can create a perfect replica of any filter using the time domain impulses, as b said. However, you won't be able to emulate the movement on the filter parameters, such as cutoff or resonance. It would work for any 'static' sound, but not for 'dynamic' ones.


    -René
    #3
    techead
    Max Output Level: -31.5 dBFS
    • Total Posts : 4353
    • Joined: 2004/01/24 08:40:20
    • Location: Macomb, IL, USA
    • Status: offline
    RE: Building Filters? 2006/11/02 22:47:55 (permalink)
    <sniff>...<sniff><sniff>...do you smell that? <sniff>...that is the tasty aroma of another tutorial baking
    #4
    Wizky
    Max Output Level: -90 dBFS
    • Total Posts : 11
    • Joined: 2005/07/07 16:33:57
    • Status: offline
    RE: Building Filters? 2006/11/03 05:35:04 (permalink)

    ORIGINAL: techead

    <sniff>...<sniff><sniff>...do you smell that? <sniff>...that is the tasty aroma of another tutorial baking


    Aye, I would vote for that too... Technically I understand b rock, but how do I implement this in practice? Sounds a bit like building convolution reverbs, and I never have had any experience with that either
    #5
    b rock
    Max Output Level: 0 dBFS
    • Total Posts : 8717
    • Joined: 2003/12/07 20:31:48
    • Location: Anytown (South of Miami), U.S.A.
    • Status: offline
    RE: Building Filters? 2006/11/03 07:49:16 (permalink)
    that is the tasty aroma of another tutorial baking
    I'm going to have to do something about that. I pride myself on being not that easy to *read*. OK, yes I do have a little time in on building the filters now. Pretty damn cool, with some unexpected results from convolving/deconvolving outside of the *norm* (including some hand-drawn IR). The best part: there are a lot of test tone impulses and IRs out there already constructed and available for download. Here's two spots: for Sound Forge from Sony, and cAPLOCK's Noisevault.
    how do I implement this in practice? Sounds a bit like building convolution reverbs, and I never have had any experience with that either
    First of all, Wizky, thanks for 'pushing' me towards loading up the V-Station and Bass Station that I had lying around; dormant and uninstalled. Neat little synths. The oscillator sync sounds great, and they respond to program changes(!).

    As for a 'tutorial', I've decided to try to work out one that uses tools that everyone would have on hand. I've got the SIR around here somewhere, and there are some freeware convolution apps as well. Maybe I can just start from available impulse responses, then point in the direction of rolling your own. I'm working off information that I gleaned from Dimension Pro, so there's some things that need to be in place for Rapture use.

    One of those things is a "silence.wav". DP has a 98 - Special folder. I don't think that Rapture ever had the same contents; I do believe that I copied it over from DP. The Impulse folder has the same file. At any rate, this "silence.wav" is nothing more than 4,640 samples of ... nothing. Digital silence. It serves as a placeholder for the 'Chain To Next Element' toggle under E1-E3 (-E5 in Rapture). You build your Elements as usual, but then you run them through the last one as a 'cabinet, amp, filter, microphone, reverb', etc. I've had limited success in writing it all to a singular .sfz file, but I'm sure it can be done.

    At any rate, here's a 'teaser'. It's taken from one of the Cabinet models, and sketches out how the time domain impulses work:

    ----------------------------------------------------------------
    <region> sample=Impulses\silence.wav

    <effect>
    type=tdfir
    tdfir_impulse=Impulses\place_your_impulse_here.wav
    tdfir_gain=0
    tdfir_dry=0
    tdfir_wet=100
    ----------------------------------------------------------------

    From what's there, you can deduce a gain compensation adjustment, and the mix of wet and dry effect. This one is 100% wet, because it's 'filtering" the Element before it in the 'Chain To Next Element' configuration.
    Edit: More as it comes.
    post edited by b rock - 2006/11/03 08:16:48
    #6
    Jump to:
    © 2024 APG vNext Commercial Version 5.1