Sine Sine Everywhere a Sine

Official support for: u-he.com
Post Reply New Topic
RELATED
PRODUCTS

Post

By using the SpectroBlend mode of the OSC module one could in theory build a progression of 16 tables that resynthesize a (short) sample by sweeping the Wave(Warp) parameter through each. My question is whether there are any examples of this in any of the Zebra patches currently available, and whether such an idea is practical vs too limited by only having 16 tables/128 harmonics. At the very least I would think that useful transients could be created this way. And speaking of the way in which this could be done, MATLAB perhaps? (I personally would use Octave because I have no money) Or some other tool(s)? Assuming some sort of raw FFT data set could be generated from an existing sample, a Zebra Script might complete the task of importing it into an OSC module. Anyway, just brainstorming...

Post

friggin geek

:hihi:

Post

Most of the brainstorming I've done here in the lab so far has been with various Sims, and here's a few quick thoughts from them:

Ha ha frattie! Oh Oh talerg larg ou ou denago. Chig dagee? Mmm meskava. Gerpla! Ha ha ha - meglerta? Chaval fifty nerchas sushbago. Erkie! Goo glabby.

Post

I regret I cannnot give you an answer, but your post title is very clever!

(I'd like to meet your Sims too!)

:)
Dave Burns
Lowell, MA

More equipment than skill...

Post

It should be possible.
I know that my blofeld have a wavetable that goes
"One-Two-three-four-five" when you sweep through it.

But how to make that kind of tables I don't know.
//L

Post

I'm beginning to think the Sims don't actually know anything about digital signal processing... I can't believe I fell under the spell of their contagious enthusiasm.

I also can't believe that I never used MATLAB (from now on, when I say "MATLAB", I really mean "Octave") for much of anything until now, but it looks viable. I can load audio files, do an FFT, and write text and numeric data directly to a file (i.e. a Zebra h2p script). So the only thing I'm still wondering is if it's worth the trouble. Damn you Italy and all your vacationy charm, luring knowledgeable people off into a seductive DSP-free world of carefree living, far far away from harmful WIFI radiation! :x

Post

A wave is magnitude/time, spectrograph is magnitude/frequency, and DFT algorithms make these fungible . . . Pretty clearly somewhere under the hood Zebra2 looks at wave and spectral OSC modes as the same thing, so I don't think there's anything possible with spectroblend that isn't possible with Jupiter8's Wav2Zebra2.

I just figured out I could control Zebrify fairly well w/ laptop microphone, which is maybe a funny kind of inverse problem - capturing pitch/amp modulations, and mapping OSC output to that. Mind considerably blown.

Post

Would spear work for the analysis part? http://www.klingbeil.com/spear/

Post

tinkerpixel wrote:Would spear work for the analysis part? http://www.klingbeil.com/spear/
What a :cool: program, thanks for the link. I have no idea (yet) how easy it would be to use its output files.

Post

I did this a while back and described it in this thread:
http://www.kvraudio.com/forum/viewtopic.php?t=250852
Don't really know why you would go the FFT route though as Z2 supports regular "sample" playback.Not that there's anything particularly wrong with that route, just playing the sample values seems easier.

Did'nt really get anything realistic out of it but it's an interesting techniqiue.

Post

I guess my question to Urs at this point is this:

Which of the following does Zebra do internally when processing a "wave warp" on a SpectroBlend set of waves? (Assume we are warping somewhere between Wave1 and Wave2)

1) Zebra converts each of Wave1 and Wave2 to time domain waveforms, and then interpolates between the two time domain data sets, producing a new time domain waveform, and plays that.

2) Zebra interpolates between Wave1 and Wave2 data sets in the frequency domain (i.e. using the SpectroBlend data), and then converts this new SpectroBlend data set into a time domain waveform, and plays that.

3) Zebra interpolates between Wave1 and Wave2 data sets in the frequency domain (i.e. using the SpectroBlend data), and then plays this data set using 128 sine wave oscillators and summing the result.

4) Zebra uses an algorithm so secret that if Urs told us, he would be forced to kill us, and that's bad for business, so don't ask.

5) Ask the Sims, they're smart enough to know the advantage of appearing stupid.

Post

billstei wrote:Which of the following does Zebra do internally when processing a "wave warp" on a SpectroBlend set of waves? (etc)
Isn't it obvious? The graphics in "more OSX" show you what's happening - simple interpolation between amplitudes.

Post

Howard wrote:
billstei wrote:Which of the following does Zebra do internally when processing a "wave warp" on a SpectroBlend set of waves? (etc)
Isn't it obvious? The graphics in "more OSX" show you what's happening - simple interpolation between amplitudes.
Then it could be either #2 or #3 above, or possibly #4 if the secret has been partially discovered, and I would only need to be killed partially, not completely.

Post

billstei wrote: Then it could be either #2 or #3 above, or possibly #4 if the secret has been partially discovered, and I would only need to be killed partially, not completely.
Maybe this is helpful? http://www.kvraudio.com/forum/viewtopic.php?t=219409

So maybe it's a little of all of #1, 2, 3 and AAAAAAGHH NINJAS

Post

billstei and xh3rv... :hihi: :tu:

Post Reply

Return to “u-he”