New G.O.S.T Analyzer plugin

Planet Z Announcements

Moderators: valis, garyb

tgstgs
Posts: 523
Joined: Sun Jan 15, 2006 4:00 pm

Re: New G.O.S.T Analyzer plugin

Post by tgstgs » Sat Jan 05, 2013 9:14 am

the values are limitted to +-max 32bit int;
----
x axis:
what about this
i make a smooth stretch variable from lin to log;
so you may choose the scale as you like

??vibes

Liquid EDGE
Posts: 866
Joined: Sun Apr 04, 2004 4:00 pm
Location: Brighton England
Contact:

Re: New G.O.S.T Analyzer plugin

Post by Liquid EDGE » Sat Jan 05, 2013 9:26 am

Hey, that would be awesome. I'm not complaining at all about how it is now (as I fully understand now that's how we hear) but to have the choice would be absolutely great.

Vibes indeed. :)

tgstgs
Posts: 523
Joined: Sun Jan 15, 2006 4:00 pm

Re: New G.O.S.T Analyzer plugin

Post by tgstgs » Sat Jan 05, 2013 10:03 am

good vibes
Attachments
TGSANAGui.zip
(711.71 KiB) Downloaded 119 times

tgstgs
Posts: 523
Joined: Sun Jan 15, 2006 4:00 pm

Re: New G.O.S.T Analyzer plugin

Post by tgstgs » Sat Jan 05, 2013 10:08 am

to be honest is no big deal course its already in;
you just have no control over it;

_save the old one;
_overwrite just the dll;
added a textfader below the dB multiply;

as well as the rightmousebutton down drag up and down over the curvedisplay sets the y multiply
a left to right drag with the rightmousebuttondown stretches the x axis choose the scale how ever you like it;

hope you like it vibes

Liquid EDGE
Posts: 866
Joined: Sun Apr 04, 2004 4:00 pm
Location: Brighton England
Contact:

Re: New G.O.S.T Analyzer plugin

Post by Liquid EDGE » Sat Jan 05, 2013 10:32 am

Now that was quick. Nice one man. Will download asap and thanks for the added choice. You are a star. :)

Liquid EDGE
Posts: 866
Joined: Sun Apr 04, 2004 4:00 pm
Location: Brighton England
Contact:

Re: New G.O.S.T Analyzer plugin

Post by Liquid EDGE » Sat Jan 05, 2013 10:43 am

Hey not meaning to be a pest. Would there be any chance you could have the y axis to be able to switch so it displays volume equally from -100 to 0 ?

So to have it as it is now with the multiplier. Or to hit a switch so it shows volume --100 to 0 with equal increments up the axis.

User avatar
dante
Posts: 3834
Joined: Sat Nov 24, 2001 4:00 pm
Location: Melbourne Australia
Contact:

Re: New G.O.S.T Analyzer plugin

Post by dante » Sat Jan 05, 2013 12:08 pm

tgstgs wrote:good vibes TGSANAGui.zip
Awesome thanks. Will try it out later.

Eanna
Posts: 602
Joined: Fri Dec 02, 2011 1:57 am
Location: Ireland

Re: New G.O.S.T Analyzer plugin

Post by Eanna » Sat Jan 05, 2013 6:53 pm

Looks like we need a definitive resource on how to use this device.
Dante, I know you started a thread that looked like it was going to create a ScopeRise article. I'd be interested in such an article!

For me, the manual doesn't explain what things are - it's functionally appropriate as a user guide for the device, but for folk like myself who don't really understand this stuff (a little knowledge is a dangerous thing!), it would be great if we could have a "what" and "why" to back up the "how" in the manual.
E.g. scenarios/use cases, what you're looking at, why you use particular settings...

I have archived a set of PDFs on Visualizers that I've threatened to read for some time... I will pull these out, print them on Monday, and get my sorry teeth into the meat of these. Hopefully, I'll be able to match what I learn in these docs to the GOST Analyzer. If I gain any insights, I'll get back on here...

For example, looking at garyb's post linking to a website describing A-weighted, C-weighted and Z-weighted graphing strategies, it would be good if the GOST graph showed dBA, dBC, or dBZ legending on the Y axis...

I do feel that this is a more fully-spec'ed analyzer than I'm currently able to attribute to it.

If anyone can contribute immediate knowledge, that would be useful...

Thanks
Not because it is easy, but because it is hard...

User avatar
dante
Posts: 3834
Joined: Sat Nov 24, 2001 4:00 pm
Location: Melbourne Australia
Contact:

Re: New G.O.S.T Analyzer plugin

Post by dante » Sat Jan 05, 2013 7:35 pm

I'm just learning as I go. And compiling our applications for a ScopeRise article. If you get some helpful documentation together I'll include it in article here :

http://www.hitfoundry.com/issue_18/ana_mast.htm

niceboy
Posts: 564
Joined: Wed Jan 26, 2011 2:51 am

Re: New G.O.S.T Analyzer plugin

Post by niceboy » Sun Jan 06, 2013 11:39 am

dante wrote:I'm just learning as I go. And compiling our applications for a ScopeRise article. If you get some helpful documentation together I'll include it in article here :

http://www.hitfoundry.com/issue_18/ana_mast.htm
Can you help me.
Where is that centering of Subbas discribed ,
I missed that and the thread is so long now .
Bear

Liquid EDGE
Posts: 866
Joined: Sun Apr 04, 2004 4:00 pm
Location: Brighton England
Contact:

Re: New G.O.S.T Analyzer plugin

Post by Liquid EDGE » Sun Jan 06, 2013 2:39 pm

Eanna wrote:
For example, looking at garyb's post linking to a website describing A-weighted, C-weighted and Z-weighted graphing strategies, it would be good if the GOST graph showed dBA, dBC, or dBZ legending on the Y axis...

I do feel that this is a more fully-spec'ed analyzer than I'm currently able to attribute to it.

If anyone can contribute immediate knowledge, that would be useful...

Thanks
GaryB just cleared this point up for me on xited.org (though he tried in this thread and i didn't get it.lol). as this is what was confusing me the most. it looks like that analyzers i'm used to looking at (like the covolveq's frequency band level meters) have a weighted effect on the levels on each frequency band to match with how our ears percieve loudness (which is why garyb posted what he posted to explain what some analyzers do). While this analyzer is just showing the truth and nothing but the absolute truth so i believe the y axis is just dB.
Last edited by Liquid EDGE on Sun Jan 06, 2013 3:01 pm, edited 1 time in total.

Eanna
Posts: 602
Joined: Fri Dec 02, 2011 1:57 am
Location: Ireland

Re: New G.O.S.T Analyzer plugin

Post by Eanna » Sun Jan 06, 2013 2:49 pm

So, do we need a mode that downscales the graph according to that curve?

Its not easy to make judgements about a mix or sub group when the analyzer isn't displaying what our ears perceive.
I appreciate that a surgical view is useful, like how a compressor might react to the signal...
Not because it is easy, but because it is hard...

Liquid EDGE
Posts: 866
Joined: Sun Apr 04, 2004 4:00 pm
Location: Brighton England
Contact:

Re: New G.O.S.T Analyzer plugin

Post by Liquid EDGE » Sun Jan 06, 2013 3:00 pm

Eanna wrote:So, do we need a mode that downscales the graph according to that curve?

Its not easy to make judgements about a mix or sub group when the analyzer isn't displaying what our ears perceive.
I appreciate that a surgical view is useful, like how a compressor might react to the signal...
I think it would be nice to have a dBa mode.

User avatar
garyb
Posts: 21690
Joined: Sun Apr 15, 2001 4:00 pm
Location: ghetto by the sea

Re: New G.O.S.T Analyzer plugin

Post by garyb » Sun Jan 06, 2013 5:22 pm

i already asked tgstgs about this. i think it's no big deal and would happen.

User avatar
dante
Posts: 3834
Joined: Sat Nov 24, 2001 4:00 pm
Location: Melbourne Australia
Contact:

Re: New G.O.S.T Analyzer plugin

Post by dante » Sun Jan 06, 2013 11:34 pm

Meanwhile, the extra scaling ability is a very nice addition :
dead-or-alive-2.gif
Wanted Dead Or Alive remix
dead-or-alive-2.gif (877.47 KiB) Viewed 3618 times

tgstgs
Posts: 523
Joined: Sun Jan 15, 2006 4:00 pm

Re: New G.O.S.T Analyzer plugin

Post by tgstgs » Mon Jan 07, 2013 1:58 am

the scale is dBfs
the multiply is done at sample base with double precision;
NOT just a curve 200 pixel multiply;
you see at full multiply that you still get highest resolution;
if you just multiply +-200 pixel it would get steped;

the same applies for the a+b values they are added at sample base in frequency domain;
if you do this in time domain you would get distortion;
wrong frequencies showing up;

as well as the frequency stretch is done at bin level;
(for the release version i must add a button to click to log view as well as the range should be optimized for all blocksize;
this controll was hidden behind the screen so there was no need to so far)

it applies for all calculations are done at the most higest resolution = double precision;
when all calculations are done the result is limitted to 32bit int scope audio format and scaled from +-2147483647 to +-200 pixel
and sent to screen;
---------

NO__ there are no weighted analyzers as far as i know;
feed with a white noise;
if you see a flat curve it is not weighted;
but i agree that it would be a nice feature to switch on;
just to have another view;
but it is not possible to do serious measurement with it;
---and
http://de.wikipedia.org/wiki/Lautst%C3%A4rke
you know that it changes with volume?

good vibes

Liquid EDGE
Posts: 866
Joined: Sun Apr 04, 2004 4:00 pm
Location: Brighton England
Contact:

Re: New G.O.S.T Analyzer plugin

Post by Liquid EDGE » Mon Jan 07, 2013 4:29 am

Ok.
I think I'm starting to understand what I'm seeing.

I really know very little about frequency analyzers and when I have used one it's been a free one and where the frequency bands are displayed differently.

For instance 1k frequency in a few analyzer pics I have looked at are in the middle while with this it's more to the left (except for the latest animated pic Dante has posted). Which is what gave me this illusion that the volume was dropping quickly going up the frequency.

I'm understanding it more now. And feel a little bit of a dimwit now. As I was and have, just purely looked at the visuals without really looking at or taking in what the x axis is displaying (frequency bands/scale).

It's starting to make more sense to me.
Last edited by Liquid EDGE on Mon Jan 07, 2013 4:48 am, edited 1 time in total.

Liquid EDGE
Posts: 866
Joined: Sun Apr 04, 2004 4:00 pm
Location: Brighton England
Contact:

Re: New G.O.S.T Analyzer plugin

Post by Liquid EDGE » Mon Jan 07, 2013 4:45 am

Anyhow's. This all strengthens why someone like tgstgs makes devices and someone like me is a user. He knows what he is doing, I don't. Lol.
Or more precisely, he's an audio scientist and I'm merely playing at being a producer/sound engineer.

tgstgs
Posts: 523
Joined: Sun Jan 15, 2006 4:00 pm

Re: New G.O.S.T Analyzer plugin

Post by tgstgs » Mon Jan 07, 2013 6:26 am

maybe this helps a bit for understanding:

filterbased analyzer:
take a lopass a hipass a bandpass;
put a VU meter at each of the filterouts;
and you have a 3 band analyzer;
you can imagine that the frequency separation depends much on the quality of the filters used;
most of them show on a lo signal outputvalues in the high vu;
course the stopband does not stop fully;
in fact convolveQ is the only scopefilter being able to drop the stopband into the non hearable area;
but still there are some values;
so low frequencys show also in the hipass and bandpass and add up there with the high contend;

fft analyzer:
using the Fast Fourier Transform you save computation time by the powers of 2;
basicaly_
audiosamples are collected up to the framesize (fft_blocksize)
than a FFT is made;
the result are (fft_blocksize/2)+1 magnitudes;
their freqeuncies (BIN frequencies) are determined by (samplerate/fft_blocksize)*binnumber
so for a fft_blocksize of 1024 at 44100 Hz you get
BIN_0 = sr/fft_blocksize = 44100/1024 = 43.06640625 Hz
BIN_1 = BIN_0 *2 = 86.1328125 Hz
BIN_2 = BIN_0 *3 = 129.19921875 Hz
...
and so on;
so the output is linear__
up to
BIN_0 * 512 = 22050 Hz = sr/2;
-------
the frequency resolution is 43.06640625 Hz

now if you stretch this linear values in a 1024 pixel display log type;
you get the left 1/3 of the display about 5 valid measurement points;
whereas on the rightmost display each pixel collects a cuple of binfrequencies;

the idea of the fix Hz scale was to show a log display up to the fundamental area
by streching the display in the high end to show more valid measurement points;
the hidden fader got a value per fft_blocksize from a logic;
the range was set to fit these needs;
now you are the logic;
it is possible to reach a full log scale at each fft_blocksize but . .
only at 1024 you have the full range from lin to log__
at 8192 as shown in the pic of dante its more a log to log overstretch__
so its no good example;
but i will make 4 different ranges for the updaterelease;
+ maybe a doubleclick for setting back to log type;
-----------

now if you take 1 bin the normal display shoes the magnitude and its phase vs the fft_blocksize;
never seen in any other analyzer btw.
you see it rotating around zero right;
if it has a fix value its phase is constant to the fft_blocksize so it must be a binfrequency;
OR you have catched the input;
now if you increase the Hz input you see it rotating - spread over 2 bins and finally showing up on the next binfrequency;
the distance is 43.06640625 Hz at 44k1 1024 frame;
if you take 2 sinus you can see its phase to each other vs the fft frame;
catch one to get their relation to each other without rotation;

this is very simplified;
i hope no scientist beats me for simplifieing;
there should be lots of info about ffts around just do a search if youre interested in;
this topic fills books you know;

no more time right now vibes

Liquid EDGE
Posts: 866
Joined: Sun Apr 04, 2004 4:00 pm
Location: Brighton England
Contact:

Re: New G.O.S.T Analyzer plugin

Post by Liquid EDGE » Mon Jan 07, 2013 6:43 am

Thanks a lot for that. Still a lot for my brain to process, think I may get a book, but yes my understanding is growing. Thanks sir.

Post Reply