Signalizer: Sidechaining update! Open-source & free audio visualization plugin (v. 0.4.3)

[emarsk]

Hi Janus, I just discovered Signalizer and it looks really good.

I was playing a bit with the spectrum analyzer, and it looks like it has a +1dB/oct slope: feeding a white noise to it, I get it to be flat with about -1dB/oct slope, and with a pink noise I need +2dB/oct. I used the JSFX generators that come with Reaper, and they look accurate when measured with other spectrum analyzers.

Also, it has a ton of parameters, but it looks like there's no configurable FFT block size, or did I miss it?

[Mayae]

Hi Emarsk,

Thanks! You can't really calibrate with white noise, unless you're averaging to infinity Using a logarithmic view, white noise looks less "dense" in the bottom (try switching to linear). Also it depends on stuff like the window function - a rectangular window function can have up to 3 dB of scalloping loss. Testing with the "flat top" window, and lanczos interpolation, I found it to be accurate to within 0.01 dB error, using a precise sine oscillator.

The block size is smoothly variable in Signalizer, it is simply the "window size" parameter.

[tapiodmitriyevich]

Just a quick "Thank you" here. Love your plugin. It is super awesome. Was always searching for a good oscillator. No more need to search now.

[emarsk]

I know that the noise spectrum looks less "dense" in the low end, that's not the issue. The point is that white noise should have - by definition! - a spectrum that looks flat on average. And it does indeed, in every other spectrum analyzer I have except Signalizer, which gives me a 1dB/oct slanted graph instead, with every window function in its list.

[Mayae]

I know that the noise spectrum looks less "dense" in the low end, that's not the issue. The point is that white noise should have - by definition! - a spectrum that looks flat on average. And it does indeed, in every other spectrum analyzer I have except Signalizer, which gives me a 1dB/oct slanted graph instead, with every window function in its list.

Well, riddle me this: If a pure sine measures completely flat, how would it be able to skew only white noise?

Joking aside, I'm sorry, but I'm simply not able to reproduce your results. Maybe provide some testcases or signals that clearly show the issue? For reference, here's what I see for white noise on a logarithmic scale:


And yes, I can see why one would say it's skewed. But white noise is not supposed to "look" flat in logarithmic views, since it distributes energy equally/linearly over the whole frequency band. So if we change the view to linear, and swap between different slopes, I hope you can see that Signalizer indeed has a flat slope by default:

[mystran]

And yes, I can see why one would say it's skewed. But white noise is not supposed to "look" flat in logarithmic views, since it distributes energy equally/linearly over the whole frequency band.

It depends on what you're plotting. Practically all "normal" FFT analysers plot straight bin-magnitudes as approximations of the peak magnitudes at various frequencies and this type of plot will result in flat line for white noise (when set for "unweighted" display, such that a sine sweep measures flat). In such a plot it's really irrelevant whether the frequency axis is set for linear or logarithmic, because the latter just places the data-points visually closer at higher frequencies.

On the other hand, an octave-band analyser would generally show flat line for pink noise (again with sine sweep measuring flat) because such an analyser tries to give you the total signal energy over frequency band, integrating over progressively larger bandwidths at higher frequencies. It's certainly possible to do octave-band style analyser using FFT internally though.

The reason why you get these two different results in various analysers is because they are simply plotting two different things (eg. amplitudes vs. energy). That said, if you don't get a flat line with either white or pink noise, then you should probably sanity check what it is that you are actually trying to plot.

[Mayae]

Hi mystran

It depends on what you're plotting. Practically all "normal" FFT analysers plot straight bin-magnitudes as approximations of the peak magnitudes at various frequencies and this type of plot will result in flat line for white noise (when set for "unweighted" display, such that a sine sweep measures flat). In such a plot it's really irrelevant whether the frequency axis is set for linear or logarithmic, because the latter just places the data-points visually closer at higher frequencies.

I think what I was previously referring to with this segment:

And yes, I can see why one would say it's skewed. But white noise is not supposed to "look" flat in logarithmic views, since it distributes energy equally/linearly over the whole frequency band.

Is that, while mathematically, averaged over infinity, such a plot should be completely flat, when you're sampling moving windows and plotting it logarithmically, it gives the illusion that the bottom is less flat, because there's less energy density that would statistically "fill" up the space, contrary to the highs where there's a larger density of bins per pixel (and the highs have quicker response due to # number of cycles over a window). So there's a higher chance that the highs more quickly converge to the expected value.

Signalizer plots the absolute values of the complex bins, sinc-interpolated when there's less than one bin for a pixel, or the max value when there's more than one bin for a pixel. For any normal example though, it is completely flat (minus the illusion) when you peak-hold the energies for a white noise signal:


And just to remove any doubt, a dirac delta impulse shows it is completely flat to within machine precision (otherwise, it is YOUR fft algorithm that's to blame ):

[mystran]

Is that, while mathematically, averaged over infinity, such a plot should be completely flat, when you're sampling moving windows and plotting it logarithmically, it gives the illusion that the bottom is less flat, because there's less energy density that would statistically "fill" up the space, contrary to the highs where there's a larger density of bins per pixel (and the highs have quicker response due to # number of cycles over a window). So there's a higher chance that the highs more quickly converge to the expected value.

I'm just trying to say that practically all FFT-type analyzers will show you a perfectly flat line, because they are plotting magnitudes, not energy density.

[Mayae]

I'm just trying to say that practically all FFT-type analyzers will show you a perfectly flat line, because they are plotting magnitudes, not energy density.

Yes, I just read your answer again and I get what you're saying.

Interestingly, when you set it up to measure something like the absolute energy in a complex resonator with a bandwidth inversely proportional to a logarithmic frequency axis (ie. a lower pixel represents a larger Q), you get a very distinct 10 dB/oct slope.

[dandezebra]

I'm just trying to say that practically all FFT-type analyzers will show you a perfectly flat line, because they are plotting magnitudes, not energy density.

Yes, I just read your answer again and I get what you're saying.

Interestingly, when you set it up to measure something like the absolute energy in a complex resonator with a bandwidth inversely proportional to a logarithmic frequency axis (ie. a lower pixel represents a larger Q), you get a very distinct 10 dB/oct slope.

Reading this makes me feel like an idiot...

Thanks for the great looking plugin! I'll give it a shot when I get some time.

[Naenyn]

How goes the development?

[brok landers]

as much as i love this plugin, for absolute knowledge on how to get the most out of it i would soooo love an in-depth manual with an in-depth decription of what every parameter is doing and what it is good for...

[Compyfox]

I second the manual, but this would be on an insane scale to make it understandable for non specialists in this field (think Christian Budde).


Other than that, I used v0.3.2 in Logic Pro X 10.3 on Yosemite in 2k resolution recently (iMac 11.3, please don't ask me for further specs - I'm not Mac savy) - trying to measure a host-bundled Plugin... Maybe it's a bug, but compared to Windows, I couldn't resize/stretch the UI. I could only push it into fullscreen mode

[hopikiva]

get to get a look at this ...windows7 64 bit and Reaper... works nicely, thanks

I would love to see in the future a way to resize the window so that we have a flexible size, even if it's a menu choice of like 100, 200, 300 % options, etc.

thanks again

[0degree]

Great stuff, but where's the donate button?