So, im a studio one user (W11 x64), with the spectrum analyzer native and others, im trying to read a song of another artist, not mine, to study where its place the kick , bass etc... i know use your ears , yess but this its another way where i feel comfortable to study another songs and try to think how was the way to arm/mix the song.

Eg: here i can say the kick its at 90-112hz picking -3.9db


But here...with AnSpec from Voxengo that 95-119hz its picking at -2.9 lets say...


Here with Spectrum Analyzer from Seven Phases its at 101hz -5db


-
So far each one are throwing very different results... i know this kind of change cause algos etc... but its very different... and lets say if i want to read this into PRO-Q 3, i dont understand why its so different if all of them are throwing the result in DB, not DBFS (dbFullScale)... why vary that much?
-

Here the kick its at 90hz lets say, but in the DB meter says its at +3DB and the meter next to, mark -24? all of this its very confusing me and cant decide HOW and WHICH use to read info the most accurately ..
its kind of driving me insane...


Thanks in advance for ANY help.

 

Tilt, window, abs and rounding.
Just use 1 spectrum analyzer. Don't drive yourself crazy!

 

sorry not following you :c

 

https://www.tek.com/de/blog/window-functions-spectrum-analyzers

Apart from that, abs must be different if the frequency bandwidths of two analyzers being compared do not match. This is in the nature of things.

 

Last edited: Jan 13, 2024

These analyzes vary in type (FFT vs 1/x octave band), amount of measured freqs (horizontal resolution), overlapping of bands, smoothing and tilt (apart from timing and vertical resolution). Use one (maybe two), learn how they work and stick to them.

 

well , i have homework to do, thanks!

 

okay, that would make my saturday, thanks!!

 

I trust fabfilters pro q analyzers most, because I know those best.
But izotopes insight might be worth checking out.

 

Looking at an audio spectrum like this won't give you much information about how to mix a song, an FFT spectrum analyzer like the ones you're using are much better at showing static audio signals that don't change quickly over time.
Listening and comparing by ear is a much more efficient way to learn about mixing, and I don't think that we have the tools yet to analyze a song in this regard efficiently. One day maybe, by the help of a dedicated AI engine.
Even if you're looking at a spectrograms with different FFT window sizes, it will be hard to derive much useful information from them about how to mix the instruments in your song and sound like your reference song.

 

hi! i forgot insight exist... maye ill check it but its not a paid one?
ye i would love to know how to read those meter at PRO-Q but i cant decide so far which one ill use x_x

 

didnt know this! , cause with the spectrum analyzer native from studio one, always let me see things! but like u said, and others comments said, theres no way any tool of these can be precise...ill just stick to one, thanks!
i was starting to lose my mind... haha how can be the same tool , bring so different results? now im kind of know.. thanks again!

 

did u know whats the meter next to, the DB meter? im confused and reading the manual doesnt help me.

 

depends what you're trying to measure and what to look at,
if horizontally flat line is supposed to be equal energy, equal loudness, equal distortion... sum of channels, which weigh curve it leans towards to, then there's tilt as well

I highly recommend this video on TDR Filters by Dan Worrall, first part shows their free analyzer (very nice, definitely try!) and then majority of video explains their new filter plugins, problems with traditional HPF/LPF and so on

 

1) There is no fixed mystical freq. for the kick. It´s always at another because you must "tune" the kick that it´s in key with your track.

2) For mixing i think a "multitrack"-Analyzer is what you searched for. Below is a screenshot from your Voxengo Span in the "Plus" Version. This is what you need to see the overlapping of the freqs. ... but you can´t really mix with this, you must use your ears sorry ;-)

Greets, Guru-Bug

 

The problem with FFT analyzers is that to achieve some acceptable precision in frequency resolution (especially in the low frequency range), you need to give them a long enough analysis window which unfortunately also reduces the resolution in time.
For example, a 2048 point FFT window size at 44.1kHz will display a mix of everything that happened during the 46 milliseconds of the analysis window, and it won't tell you if it's a steady tone or a mix of drum hits or whatever. And then you only have about 22Hz resolution.
From a mathematical view, the FFT does deliver that information in the phase response but that isn't usually shown in the display because I know of no human being who is able to decipher anything useful from it

 

It's a subjective tool. It is an indication. But an indication only helps if you know how to use it for yourself. People often see it as an objective tool. It is not.

You need a reason to configure it like you do, because you know what to get out of it on a daily repetitive manner.
The same goes for all of this fancy metering by the way.

 

Last edited: Jan 14, 2024

dB @
(
90+91+92+93+94+95+96+97+98+99
+100+101+102+103+104+105+106+107+108+109
+110+111+112
) Hz
/ 23
= x dB

dB @
(
95+96+97+98+99
+100+101+102+103+104+105+106+107+108+109
+110+111+112+113+114+115+116+117+118+119
) Hz
/ 25
= x dB


Do you see your mistake?
Aside from the fact that many other factors already mentioned will skew the results of the comparison, this alone makes it impossible to compare the values.

 

Last edited: Jan 14, 2024

? Yellow refers to the bands, light grey to the signal and the meter shows peak.

 

Mkey, maybe I need to overthink my live choices.
Now I did a little comparison between fabfilter pro q and more common spectral analyzers and found that how pro q does it is not how I want it to look. I never used it for overall f-plots, just to find what to eq.
That pic shows pink noise fed to both steinbergs supervision and pro q. There is no right or wrong, fabfilter weighs non logarithmic I guess. But then again, human frequency sensing is logarithmic, so.. lets say pro q does not how human ears do it.

 

Plus, at a sample size of 2048 (at 44.1kHz), any waveforms longer than 46ms can't be accurately measured in the window, so any frequencies displayed below 217Hz are just guesswork.