KVR Audio

PeterL · Post by **PeterL** » Mon Sep 26, 2005 3:34 pm

alex.g wrote:
When you load the plugin WL comes up with an error, although the plugin works and shows the spectrum.
Would you be so kind to retry, and to email me the error at info@ag-works.net?
I'd prefer to keep these testing information out of the forum, not to pollute it

Yes, but you will have to wait until tomorrow, because I'm not on my DAW at the moment.
Maybe another user could e-mail you earlier.

alex.g · Post by **alex.g** » Mon Sep 26, 2005 7:40 pm

PeterL wrote:When you load the plugin WL comes up with an error.

I think I've fixed the wavelab problem.
WL doesn't answer to GetTimeInfo(), crashing sg-1 when it tries to get transport position.
Can I send you by mail this new version to test, before uploading on the website?

alex.g · Post by **alex.g** » Mon Sep 26, 2005 7:56 pm

Kingston wrote:Hi,
the chorus sounds beautiful. It's a very similar sound to the Univibe (the legendary stompbox) vibrato mode.
With this approach, do you think you might be able to add the "phaser mode" of univibe to your plugin as well?
(I can provide you with the schematics of the original univibe if that's of any help)

Hi Kingston,
I'm glad you're satisfied with ch-1 (although it still misses a feedback parameter and a gui..)!
Unfortunately I only heard about the univibe, never used it... anyway, I'd really like to see these schematics, perhaps it's possible to work on them.
I can't promise nothing, but I really like this kind of effects, so why not to try

Let's see if I can find some univibe sound samples now...

helium · Post by **helium** » Wed Sep 28, 2005 7:56 am

When I use your sonogram I get this:

A serious error was caused by the plugin "Sonogram SG-1". WaveLab can't work faultless nder this circumstances. bla bla

Paulie Phonick · Post by **Paulie Phonick** » Wed Sep 28, 2005 11:11 am

Thanks for sharing those two plugs, alex.g! CH-1 sounds great and the SG-1 is a really useful tool! Since both of the run without any problems in eXT, I'll just drop a couple of FRs for the SG-1:

1. Zoom level beyond 16 (64 maybe) [edit] and higher FFT resolutions[/edit]
2. An option to have highlighted lines on the sonogram for main ruler divisions
3. Note display for onclick measurements in addition to Hz

That's what came to my mind while watching what CH-1 does to my guitar output

autloc · Post by **autloc** » Wed Sep 28, 2005 4:14 pm

Nice plugins - and I really like the analysis you provide on your website comparing your chorus to conventional linear-interpolation choruses. What order of allpass interpolator do you use? Rather than the allpass filters' flat magnitude response, I would think that it's main advantage (vs. higher order Lagrange FIR interpolators) in a chorus might be that it would introduce some dispersion, making the comb filters of the delay lines non-harmonic and thereby reducing the flanging as phases cancel. In general, I thought that the lowpass characteristic of linear/Lagrange interpolators was desirable given that it offers some suppression of transients/interpolation noise.

alex.g · Post by **alex.g** » Wed Sep 28, 2005 5:19 pm

Hi Paulie,
thank you for your feedback!

Paulie Phonick wrote:1. Zoom level beyond 16 (64 maybe) and higher FFT resolutions

At a zoom ratio of 8x, you actually have the relation 1 pixel = 1 FFT bin.
A 16x zoom is already interpolated at 8192 fftsize.
If you try feeding a pure sine into sg-1, for example a 5kHz sine, you'll see highlighted the frequencies between 4990 and 5010 (more or less, don't remember): this means that by now, with this fft and windowing properties, you can't obtain a great detail over 16x (I did some experiment with a 16384 fft / 100% windowing and this rule still applies). It could be an idea when, in sg-2, FFT properties will be parametric. Anyway, do you really need a 32768 or greater FFT?

2. An option to have highlighted lines on the sonogram for main ruler divisions

Good idea, I'll append it to my TODO list

3. Note display for onclick measurements in addition to Hz

I already thought about this, it should be written in the "future development" section in plugin's page, anyway thanks for highlighting this thing. In particular, it would be a nice feature of sg-2, which should support a log frequency scale representation.
[EDIT]You're right, it wasn't explained that well on the website: I'll change it to "Note names near to frequencies, and during scrub"

[/EDIT]

That's what came to my mind while watching what CH-1 does to my guitar output

Guitarz rule!

Greetings,
Alex

rockstar_not · Post by **rockstar_not** » Wed Sep 28, 2005 5:36 pm

Sonogram works fine in Tracktion. I've got to agree to the Logarithmic axis on the frequency display as being a much needed improvement or option.

-Scott

alex.g · Post by **alex.g** » Wed Sep 28, 2005 9:25 pm

autloc wrote:Nice plugins - and I really like the analysis you provide on your website comparing your chorus to conventional linear-interpolation choruses. What order of allpass interpolator do you use? Rather than the allpass filters' flat magnitude response, I would think that it's main advantage (vs. higher order Lagrange FIR interpolators) in a chorus might be that it would introduce some dispersion, making the comb filters of the delay lines non-harmonic and thereby reducing the flanging as phases cancel. In general, I thought that the lowpass characteristic of linear/Lagrange interpolators was desirable given that it offers some suppression of transients/interpolation noise.

Hi autloc,
I couldn't have explained it better

Your observation about all-pass dispersion is truthful: we can notice the lack of harmonics you're talking about, from "figure 2" in the plugin's page. Linear interpolation reduces high frequencies and also introduces a flutter, which is easily audible by non-trained ears too (I experimented with some friends

)...
I used a first order all-pass, with time varying filter coefficients of course, which depend on the fractionary part of the index of the sample to be output. This fractionary index comes out from the lfo oscillator which trottles back and forward in the buffer. The filter coefficients are warped, to reduce phase distortion.
I think that (particularly with guitars and voices), sound clarity is very important, and perhaps it's not good to reduce a good signal's presence to mask some artefacts we add.
Thank you for pointing out these interesting facets.

alex.g · Post by **alex.g** » Wed Sep 28, 2005 9:37 pm

By the way,

Wavelab users, stay tuned because I think that for tomorrow we could have that bug fixed

Just a question: anyone has noticed some of his plugins running "stroboscopic" (non-smooth) or in a certain manner with the gui delayed in wavelab?

I'm referring to WL version 4.

Paulie Phonick · Post by **Paulie Phonick** » Wed Sep 28, 2005 10:06 pm

alex.g wrote:It could be an idea when, in sg-2, FFT properties will be parametric.

Cool! Parametric FFT params! That will do

alex.g wrote:Anyway, do you really need a 32768 or greater FFT?

Higher FFT sizes and zoom levels can be useful for analyzing bass parts. 8192 hardly gives enough detail to see what's played in the low end of a 5-string bass, while at 32768 you can easily see what's going on.

Anyway, can't wait for SG-2! Will there be a CH-2 as well? (with a GUI)

rockstar_not · Post by **rockstar_not** » Thu Sep 29, 2005 2:51 am

Paulie Phonick wrote:
alex.g wrote:It could be an idea when, in sg-2, FFT properties will be parametric.
Cool! Parametric FFT params! That will do

alex.g wrote:Anyway, do you really need a 32768 or greater FFT?
Higher FFT sizes and zoom levels can be useful for analyzing bass parts. 8192 hardly gives enough detail to see what's played in the low end of a 5-string bass, while at 32768 you can easily see what's going on.

Anyway, can't wait for SG-2! Will there be a CH-2 as well? (with a GUI)

here's an idea:

Software that I use at work to analyze automotive sounds uses a variable FFT size to analyze different parts of the spectrum so that 'visually' the spectral lines look similar in width on a logarithmic scale.

So, bass end of the spectrum gets a 'zoom' fft with high number of points, while higher end of the spectrum uses less number of fft points in the windowing.

Is that possible with your code?

-Scott

Paulie Phonick · Post by **Paulie Phonick** » Thu Sep 29, 2005 6:55 am

Are you sure this would be more optimal (CPU-wise)? After all it would require calculating a couple of FFTs of the same signal at various sizes. Doing one higher size FFT and scaling the results should be faster imho. On the other hand having spectral lines of similar width on the log scale could be a nice feature so why not have a variable FFT size mode as well?

helium · Post by **helium** » Thu Sep 29, 2005 10:14 am

You could downsample the signal and use this lower resolution signal for the detailed bass FFT. This way you'd get a pretty high bass resolution with a small FFT size.

PeterL · Post by **PeterL** » Thu Sep 29, 2005 10:20 am

alex.g wrote:Wavelab users, stay tuned because I think that for tomorrow we could have that bug fixed

Just a question: anyone has noticed some of his plugins running "stroboscopic" (non-smooth) or in a certain manner with the gui delayed in wavelab? I'm referring to WL version 4.

I've tested your new version. No problems at all in WL4.
I think you could put it online.
I sent you an e-mail with detailed results.

Some new plugins to test...