Any free spectrograph vst (or app)?

VST, AU, AAX, CLAP, etc. Plugin Virtual Effects Discussion
RELATED
PRODUCTS

Post

jmh wrote:Wow, 3D just got redefined. Geez, one can learn a lot from this forum...
JMH
No 3D is already well defined.
But a 3D FFT analysis is usually even a 3D graphic, not a two dimensional waterfall diagram. :P

Because FFT is conceptually already "3D". Frequency, Amplitude, Time.

.

Post

:)

I wasn't referring to what you said though, I'm agreeing with you totally on this one (as usual! :P)

It's the first time I've seen anyone seriously trying to explain that a waterfall graph is 3D =)

Stretching existing terms is something that is getting more and more popular here, aint't it :)

* edit - waterfall, not faterfall *

JMH
Now available with added Inherently Suspect Justification!

Post

OOO,

The spectrogram that was originally shown is what seems most automotive sound quality specialists prefer for one main reason: Nothing is 'hidden' by tall peaks in the foreground like the waterfall plot that you have shown.

Of course we do have this kind of a plot available for impressing management when such is necessary, but there is absolutely no additional information presented in the example plot that you provided than what was in the original 3D plot given.

I would take back this argument if somehow there was a virtual space (placement in the L/R, Fore/Aft axes) dimension somehow analyzed and displayed but that does not seem to be the case. In fact the plot that you show seems to put color in the plot just for fun - it's not showing any new information than the same thing as the position along one of the axes. Normally, with waterfall plots that I have seen, color is used in conjunction with vertical position to indicate amplitude.

Make no mistake, 'flat' spectrograms that represent 3 dimensions like the original picture, are very often referred to as 3D plots, not only in sound analysis but in other fields where such plots are used as well.

-Scott

-Scott

Post

000 wrote:
jmh wrote:Wow, 3D just got redefined. Geez, one can learn a lot from this forum...
JMH
No 3D is already well defined.
But a 3D FFT analysis is usually even a 3D graphic, not a two dimensional waterfall diagram. :P

Because FFT is conceptually already "3D". Frequency, Amplitude, Time.

.
Here I must take exception again to terminology that you are using.

The spectrogram originally shown is not a waterfall plot. It's usually referred to as a spectrogram or spectrograph as the original poster called it.

A 'waterfall' plot, is called such because it can have the same appearance as a waterfall - not unlike the picture that you showed.

Regarding traditional FFT analysis, outside of the music world, most folks don't look at FFT vs. a time axis when dealing with more steady state sounds (not in music but in other areas of science) They typically look at an 'averaged' plot which is in fact just two dimensions - with frequency along the x axis (horizontal axis) and amplitude displayed on the y (vertical axis). Voxengo's SPAN is just one such example that you can pick up for free through KVR.

-Scott

Post

Out of curiosity, trying to come up with a conclusion on how these terms are used, I went and did some googling...

3D plot results in "real" 3D plots, that is, those typical 3D pictures most of us understand as 3D images.

Spectrogram results in spectrograms - where three different attributes are presented in a flat image, which most people don't call 3D.

Waterfall graph results in... lo and behold, an equal amount of both, as if there wasn't a common meaning for the term.

In any case, coming from a graphics background, I tend to consider 3D as a term that is almost universally understood as 'three dimensional images' - the stuff most people come up with in rendering applications...

So, no wonder one commonly understood term is being discussed, as apparently there isn't a consensus of its meaning in the outside world.

Regards,

JMH
Now available with added Inherently Suspect Justification!

Post

Also, worth adding I guess, is that in 3D modeling world there are height/bump/displacement maps that contain 3 attributes just like spectrograms - position information and height. Never seen anyone refer to these as 3D images.

Regards,

JMH
Now available with added Inherently Suspect Justification!

Post

Well, my background is from sound analysis - first learning the terminology by trade, and then later on, as part of a work-study program, getting a Master of Science in Engineering degree from Purdue University where I took EE638; Digital Signal Processing. Spectrograms represent 3 unique dimensions of data; can be presented both as flat and faked 3 dimensional displays, but nearly always preferred as flat displays for reasons I already mentioned.

Waterfall terminology came from the kind of line-plot that was possible on line plotters where a sort of wire-frame was presented - usually with just one color.

Here are some google results - or just do it for yourself by typing in 'spectrogram examples'
http://en.wikipedia.org/wiki/Spectrogram
http://dsp.ucsd.edu/students/present-st ... ecgram.htm
http://www-users.cs.york.ac.uk/~alistai ... mplex.html

I guess I can't find an example of a waterfall plot that is a flat spectrogram display.

Here is the help file from MATLAB (probably the most universally used data analysis package on the planet) on what a Waterfall plot is according to them.

http://www.mathworks.com/access/helpdes ... rfall.html

So, no, a spectrogram is not a 3D image, but it is a way to display 3 dimensions of data in one display and folks that use them will refer to them as 3D displays. And again, these are almost universally used in publication no matter if it's sound analysis, chemical analysis, and so forth, because there is no 'hidden' data, as one can run into with waterfall plots. Waterfall plots are nice to interact with on a computer if you have 'spin' capability on the axes (spectrograms are just a waterfall plot viewed along the vertical axis) but without that capability, one almost always runs the risk of not seeing all the data because of tall peaks in the foreground.

If you'd like more examples of papers where spectrograms are referred to as 3D displays, then I would point you to the archives of the Acoustical Society of America and it's journals, the Society of Automotive Engineers Noise & Vibration conference proceedings from the past 20 years, etc.

-Scott

Post

Hey guys, I said 3D to contrast it against 2D spectral analyzers like the ones from Voxengo,etc.
In fact, I find that using colors to reflect amplitude is much better than using a 3rd axis, it's much clearer to read.
But anyways, do you guys have any to recommend?

QSpec Pro looks very complete, but it adds a strange shadow to the spectrogram, making it almost unusable. And it also has a 3-axis mode, but it's just too small to be usable.

Andrés

Post

Audacity standalone has it as mentioned back on page one. Not a plugin, but decent to have in the toolbox.

-Scott

Post

Scott Thanks, this was once again learning experience :) Out of curiosity, I called a friend who does all kinds of scientific analysis and asked about this, and got an answer I suspected - in finnish (which is my native language) we'd never call them 3D images, but in english it's a common term indeed.

So, I stand corrected, twice :)

Regards,

JMH
Now available with added Inherently Suspect Justification!

Post

Interesting thread. Regardless of what one wants to call it, how is it actually used. When I look at the plot I can understand that three components are being represented but beyond that I get pretty lost? What does it actually tell me and how could/would I use the information it provides?

Post

jmh wrote:Scott Thanks, this was once again learning experience :) Out of curiosity, I called a friend who does all kinds of scientific analysis and asked about this, and got an answer I suspected - in finnish (which is my native language) we'd never call them 3D images, but in english it's a common term indeed.

So, I stand corrected, twice :)

Regards,

JMH
I would never call it a 3D image either, but it's a data display that properly displays 3 unique dimensions of data.

By the way, this guy wouldn't be Antti Järvinen would it? He was a guy I met at a conference on Sound Quality from Technical University of Helsinki. He was doing binaural research from a safety perspective on these huge logging machines, where the operator of the machine must rely on his hearing skills and ability to identify directional sounds. It was quite interesting work. At that time, if I recall correctly, he claimed that just about everyone going through a signal processing regimen at HUT was nearly guaranteed a decent job at Nokia.

Anyhow, answering Darcy's question....

These kinds of displays can be useful to help in pitch correction software. In reality, I find 2D (just frequency and amplitude) more useful for music analysis because the time dimension is always changing (rhythm) in the kinds of music I like. I don't really need the analysis to tell me that the content is changing in the time dimension, but it's always handy to see if there is a room resonance that may need taming, or other offensive 'spike' in the frequency data that may not be desirable.

-Scott

Post

darcyb62 wrote:Interesting thread. Regardless of what one wants to call it, how is it actually used. When I look at the plot I can understand that three components are being represented but beyond that I get pretty lost? What does it actually tell me and how could/would I use the information it provides?
Well. It enables you to examine the frequency components over the time. Extracting the pitch of the (sine) partials and their amplitude. (As you may know, any audio signal can be seen as a stack or mix of many sine tones with differing phases - which can be extracted and thus be analysed with FFT).

That is finally, that you are able to see what frequency compinents are in the analysis material at a certain point of time and which amplitude each of them have.

With good anylysers you are able to edit the colors. But normally 2 colors are enuff to display that information satisfactory (bicolor). For some reasons humans tend to use "fancy" colors ...


If you do that analysis with a singer for instance, you can clearly see in what pitch and loudness the singer performs. (It looks then quite like a blurred MIDI key roll ...)


That has currently very few to do with "pitch correction" as mentioned above (and I don't know any such pitch corrector applications, which performs such an diagram, by the way). It is rather a kinda "pitch tracking" of the *entire* frequency spectrum for pure analysis purposes - you see not merely one ( the fundamental ) frequency - which merely would be any usefull with pitch correctors finally.

Pitch correction would also require to extract the main frequency form the entire analysis and then scale the final result regarding a comparsion to a chromatic frequency table ...

But there are other (faster and more efficient) techniques for pitch detection available (and used) than FFT. FFT also *always* introduces audible latency the higher the resolution (analysis window size) is. Too short analysis window sizes (has nothing to do with the displaying windows in the operating system) are not able to to analyse low frequency content satisfactory, too long windows average the results too much, by the way ...

.

Post

000 wrote:
That has currently very few to do with "pitch correction" as mentioned above (and I don't know any such pitch corrector applications, which performs such an diagram, by the way). It is rather a kinda "pitch tracking" of the *entire* frequency spectrum for pure analysis purposes - you see not merely one ( the fundamental ) frequency - which merely would be any usefull with pitch correctors finally.

Pitch correction would also require to extract the main frequency form the entire analysis and then scale the final result regarding a comparsion to a chromatic frequency table
I was thinking of the pitch display from Melodyne see screenshots here where melodyne has already resolved things to the fundamental frequency and plotted according to scale note rather than frequency.

But even then, that is a stretch. I don't think it's too far of a stretch to make the claim that most folks do better EQ work with simple frequency/level plots, either by FFT or by 1/3 octaves from digital filters, or some other frequency/level display.

-Scott

Post

I like to use it to see the timbre of "simple" non harmonic components, like kick drums or snares. It doesn't serve as well for pitch changing instruments in my opinion.

Anybody used QSpec Pro, and know how to get rid of that "shadow" (looks like interleaved video) that displays over the spectrogram?

Post Reply

Return to “Effects”