i was begging for that option... hopefully he'll do that. would make the best even better.Massimof wrote:and... can we hope, sooner or later, in a 1/12 octave resolution?
Would be reaaaally a great feature, IMHO
Seven Phases Spectrum Analyzer
-
- KVRAF
- 6168 posts since 26 Sep, 2003 from right here, as you can see ...
regards,
brok landers
BIGTONEsounddesign
gear is as good as the innovation behind it-the man
brok landers
BIGTONEsounddesign
gear is as good as the innovation behind it-the man
- KVRist
- Topic Starter
- 347 posts since 20 Apr, 2005 from Moscow, Russian Federation
In v1.09 ;)zeep wrote:Is there a way to remember my settings as default?
_
brok landers, Massimof
To be honest, no, I do not plan the 1/12 feature in the foreseeable future (just because it's a complicated story for the moment).
-
- KVRAF
- 1794 posts since 17 May, 2005
Very nice! Thanks again for making this all free of charge.Max M. wrote:In v1.09zeep wrote:Is there a way to remember my settings as default?
Now the only thing that i can possibly wish for is a snapshot functionality that allows me to overlay an 'ideal' curve. Have you given that thought?
Oh us free plugin users, we keep asking for more!
- KVRist
- Topic Starter
- 347 posts since 20 Apr, 2005 from Moscow, Russian Federation
Maybe. But for sure not too soon.zeep wrote:.. a snapshot functionality ...
-
- KVRAF
- 6168 posts since 26 Sep, 2003 from right here, as you can see ...
ah, ok, i see... a shame... but thanks anyway - it's imo still the best analyser out there.Max M. wrote:>snip< To be honest, no, I do not plan the 1/12 feature in the foreseeable future (just because it's a complicated story for the moment).
regards,
brok landers
BIGTONEsounddesign
gear is as good as the innovation behind it-the man
brok landers
BIGTONEsounddesign
gear is as good as the innovation behind it-the man
-
- KVRAF
- 2548 posts since 13 Mar, 2004
edit:
My error, mystran's explaination is a few posts down.
====
Late to the party but checking out this analyzer currently, thanks for providing it for free.
Running pink noise through it I get a straight line when slope is set to 0, but actually I thought I should get the straight line when slope is set to 3 (as in Span) as pink noise falls of 3 db / oct.
Or am I missing / confusing something here ?
My error, mystran's explaination is a few posts down.
====
Late to the party but checking out this analyzer currently, thanks for providing it for free.
I'm testing this (running currently latest v1.09 beta) and wondering - is the slope setting implemented correctly ?Max M. wrote: Spectrum Analyzer v1.07-alpha update is available.
Changes:
- new: average slope setting
Running pink noise through it I get a straight line when slope is set to 0, but actually I thought I should get the straight line when slope is set to 3 (as in Span) as pink noise falls of 3 db / oct.
Or am I missing / confusing something here ?
Last edited by No_Use on Tue Jul 28, 2015 1:23 am, edited 1 time in total.
-
- KVRAF
- 6168 posts since 26 Sep, 2003 from right here, as you can see ...
that's called "pink noise flat", and it's beneficial for a lot of uses. but ideally the wheighting could be userajustable, which is so in some other analysers. but i love the "pink noise flat" wheightig, as for mastering it's beneficial, as the high frequency content is more visible...
regards,
brok landers
BIGTONEsounddesign
gear is as good as the innovation behind it-the man
brok landers
BIGTONEsounddesign
gear is as good as the innovation behind it-the man
- KVRAF
- 7888 posts since 12 Feb, 2006 from Helsinki, Finland
The spectral magnitude of pink noise falls 3dB/octave, but there is equal energy in each octave (or other equal sub-bands on log-freq scale). This makes sense when you consider that higher octaves cover larger frequency ranges so there are more individual frequency contributing.No_Use wrote: Running pink noise through it I get a straight line when slope is set to 0, but actually I thought I should get the straight line when slope is set to 3 (as in Span) as pink noise falls of 3 db / oct.
[...]
Or am I missing / confusing something here ?
With a "regular" FFT analyzer, you normally plot the peak amplitude around different frequencies. This results in a 3dB/octave decay with pink noise. With an "N-per-octave" analyzer, you normally plot the total energy in each of the analysis band. This gives you a flat result with pink-noise.
So for typical signals (spectrally smooth and dense), these two types normally different by 3dB/octave, but it's not really a matter of different weighting as such. Rather they are measuring (and then drawing) different information.
So I suspect what's going on here is that you are looking at an N-per-octave energy plot, when you are expecting an amplitude plot.
- KVRAF
- 7888 posts since 12 Feb, 2006 from Helsinki, Finland
Yeah, just realised the source is available, so sanity checked what it does and it runs a parallel filter bank. So each bank will basically show the total energy over that range of frequencies, which gives a flat result with pink noise sort-of naturally.No_Use wrote:mystran, very informative post, I indeed wasn't aware of the different types of analysers here.
For extra fun, try running a slow sine-sweep through both types and see what happens.
- KVRist
- Topic Starter
- 347 posts since 20 Apr, 2005 from Moscow, Russian Federation
Yep, I'm just confirming what mystran wrote. (Except that I would not use "peak-amplitude-vs-energy" words in this case as potentially misleading (as those maybe confused with timing behaviour of "peak-vs-amplitude-vs-average-vs-energy" stuff)... It's just about frequency scale these thing work: FFT works in linear frequency domain, a filter-bank of an analyser works in logarithmic domain, thus the difference is how much of signal each "point/band" gets. In simple words, for logarithmic (e.g. "per octave") bands the higher band is getting signal from a larger frequency range, thus the natural "flat line" is pink-noise)
In other words, to translate the slope setting of SPAN to the slope setting of this plugin simply substract 3dB, e.g.:
3dB -> 0dB, 0dB -> -3dB etc.
Technically, it's actually valid to think of this as just of different "slope calibration", and the "information" they show in their "average" plots (when the settings match) is equal <- upd: not quite true (see the post below) ... (for "static" signals, note however they have vastly different ballistics behaviour, roughly speaking: SPSA "average" uses rectangular-window RMS, while SPAN "average" is formed by "magnitude of signal in a FFT-window postprocessed (if I recall correctly) with VU-like ballistics", thus the result will never be the same for a "non-static" input).
In other words, to translate the slope setting of SPAN to the slope setting of this plugin simply substract 3dB, e.g.:
3dB -> 0dB, 0dB -> -3dB etc.
Technically, it's actually valid to think of this as just of different "slope calibration", and the "information" they show in their "average" plots (when the settings match) is equal <- upd: not quite true (see the post below) ... (for "static" signals, note however they have vastly different ballistics behaviour, roughly speaking: SPSA "average" uses rectangular-window RMS, while SPAN "average" is formed by "magnitude of signal in a FFT-window postprocessed (if I recall correctly) with VU-like ballistics", thus the result will never be the same for a "non-static" input).
Last edited by Max M. on Thu Jul 30, 2015 8:26 pm, edited 2 times in total.
- KVRAF
- 7888 posts since 12 Feb, 2006 from Helsinki, Finland
I mostly don't disagree with what you said, but I wouldn't call it just a different slope calibration.Max M. wrote:Technically, it's actually valid to think of this as just of different "slope calibration"
Both types of analysers (ie. uniform spacing in linear vs. log-scale) are normally calibrated to show a flat result in response to a slow, constant amplitude sine-sweep (and this also appears to be the case for SPSA). So in terms of input consisting of single frequency, the response "slope" is actually exactly the same.
At the same time, when the input is pink (or white, or whatever) noise, there is a 3dB/octave difference in measured slope, since the per-octave bank adds together progressively larger bandwidth (in the linear-scale sense) for higher frequencies where as a linear filter bank (ie. FFT) doesn't.
For most musical signals with harmonic overtones, the difference varies smooth between these two extremes with low harmonics (assuming sufficient resolution) measuring individually with the same slope in either case and high harmonics measuring with a 3dB difference.
These differences can certainly be removed by further processing if desired (eg. by correlating the FFT spectrum with per-octave FIR kernels, or using an averaging scheme that divides by bandwidth rather than band counts), but without some post-processing the difference is certainly more than just a simple slope tilt.