User waveforms

[kodama]

For a nice quality I'd see something different from just filling up 16 x 128 samples/harmonics as in the drawable waveforms. I'd rather say 2048 samples per cycle and some kind of meta controls to enhance the waveform.

The algorithm should take .wav files (and other formats such as .aiff, .sd2) of arbitrary length, find zero crossings, resample them, maybe do optional formant shifting (if I get that baked) and may even work out some grainy scheme if desired.

That pretty much involves many technologies that are also required for sampling.

The playback is the smallest part here, as Zebra's oscillators are basically dynamic resynthesizers of the waveform/spectrum that's drawn.

And of course, the gui framework has to be enhanced for drag'n'drop, stuff like that. There's a whole bunch of logistic considerations behind that, and it has to be designed to work in a universal way for all kinds of similar applications.

We weren't asking for all that! We just wanted something functional sooner than perfection brought to us by the ghost of Christmas future.

I think you may have a bit of over-achievers syndrome (or OCD)


It's like if I said "hey can I borrow a dollar"?

And you said "Cool, but I'm gonna invent a new type of currency made from a new metal that puts other curruncies to shame, but that will have to be after I start my new island nation that surpasses all other nations".

And I would be like "ummm I just wanted to buy a soda, and I don't think they take Urs-Marks!".

[Jeremy_NSL]

kodama

Gotta agree with you there, Urs doesn't do anything halfway

[Urs]

Well, you might be right... but on the other hand I think that just importing waveforms to the 128 slot GeoBlend would probably be a bit disappointing...

[kodama]

Understood! (Although I think you're just being too hard on yourself, Sytrus only had 128 slots until recently)

A pop-up menu of directories of waveforms and msegs would actually be more helpful in the meantime.

Having to create or copy/paste waveforms and msegs really stumps me when I want to start building patches in Z2....

[srcodling]

I kind of agree with kodama... having a set of pre-drawn waveforms to choose from (and then edit/mangle) would be a great addition for me... or am I just missing how easy it is to do myself???

[Kebmaster]

I have to agree. Preset waveforms would be a great addition- I'm not really interested in drawing waveforms-just tweaking them and applying waveform fx to them.

[replicator]

Well, you might be right... but on the other hand I think that just importing waveforms to the 128 slot GeoBlend would probably be a bit disappointing...

I have sampled some single cycle waveforms from an analog synth, just to try how it will sound in a plugin. The result was actually a bit disappointing when comparing the sound with the original hardware. The sampled waveform sounded dull, and not fully rich in harmonics as the original. It is true that the sampled wave was in mono but, a stereo file will just double the sample anyway.

One interesting thing I notized was that the sampled waveforms had a quite special shape compared to a classic saw, sine and so on.

Have other users experienced the same? Are there some special techniques involved to clone an oscillator?

Best regards

Replicator

[Howard]

...the sampled waveforms had a quite special shape compared to a classic saw, sine and so on. Have other users experienced the same?

Interesting. I've just been sampling Little Phatty waveforms to see what they look like - with very unexpected results. Maybe setting a different phase for partials dramatically changes what a wave looks like, while hardly changing the sound at all. Any other possible explanations?

[beej]

Yeah, oscillator waveforms, especially when generated by analog electronics are often quite far from the mathemaical "ideal" of what a sawtooth/square etc should be.

I think it's both a combination of natural characteristics of some of the components they used, plus some tweaking by ear to be where the designer thought they sounded nice - after all, the most important factor is how the sound, rather than how they look...

[zircon]

Howard; take this for what it's worth, but in one of the courses I'm taking here at school for a degree in music tech & business, we learned that the phase of partials indeed does not seem to affect our perception of sound significantly. Intensity seems to be the primary factor.

[pdxindy]

Yeah, oscillator waveforms, especially when generated by analog electronics are often quite far from the mathemaical "ideal" of what a sawtooth/square etc should be.

I think it's both a combination of natural characteristics of some of the components they used, plus some tweaking by ear to be where the designer thought they sounded nice - after all, the most important factor is how the sound, rather than how they look...

I've been learning about waveforms and sound by drawing them in Zebra. Little tweaks can make a big difference, or sometimes little difference. And yeah, listening is what is important...

[FEV]

For a nice quality I'd see something different from just filling up 16 x 128 samples/harmonics as in the drawable waveforms. I'd rather say 2048 samples per cycle and some kind of meta controls to enhance the waveform.

The algorithm should take .wav files (and other formats such as .aiff, .sd2) of arbitrary length, find zero crossings, resample them, maybe do optional formant shifting (if I get that baked) and may even work out some grainy scheme if desired.

That pretty much involves many technologies that are also required for sampling.

The playback is the smallest part here, as Zebra's oscillators are basically dynamic resynthesizers of the waveform/spectrum that's drawn.

And of course, the gui framework has to be enhanced for drag'n'drop, stuff like that. There's a whole bunch of logistic considerations behind that, and it has to be designed to work in a universal way for all kinds of similar applications.

We weren't asking for all that! We just wanted something functional sooner than perfection brought to us by the ghost of Christmas future.

As I'm a little bored at work right now , so I produced a tiny python script, that should convert 16-bit mono wave files into Z2 osc presets. Grab it from here: LINK
Obviously you'll have to install python first.

Script takes only the first 128 samples from a wave file (zero padding is performed when there's not enough samples), so you'll propably need to do some sample preprocessing to "time squezee" your input waveforms. Read the script header for some more info (usage).

As I'm at work now (no soundcard, no zebra, no nothing), I can't guarantee that it'll actually do the job

cheers,
Bart

[raikard233]

That's very interisting!
Thank you FEV!

Rik

[replicator]

Thanks Fev, I will try it

Replicator

[Howard]

Script takes only the first 128 samples from a wave file (zero padding is performed when there's not enough samples), so you'll propably need to do some sample preprocessing to "time squezee" your input waveforms. Read the script header for some more info (usage).

Thanks Bart! Er... I don't want to read the entire Python documentation for this - what's the precise syntax in the command line if I e.g. put your script and "test.wav" in Python's root directory, and want it to build "test.h2p"? I'm always getting the "invalid syntax" message.

BTW: I'm PC/Windows

Cheers!