Spectrum Analysis Problem (Resolved)

[dpaterson]

Hi.

Just when I was about to try something (mixing using pink noise as a reference) things got complicated right from the get go (as per usual).

Take a look at the attached screenshot or at the video (https://youtu.be/QAt1Y7M0gBo).

Why do three different spectrum analysers show different results for the exact same sample??? Is this something to do with plugin integration or am I missing something here???

Basically: I downloaded a pink noise sample and added it to three different tracks in Mixcraft. On each track I added a different spectrum analyser (iZotope, Voxengo SPAN Plus, and Fab Filter Pro-Q 2). Voxengo and Fab Filter appear to be in agreement with each other but iZotope shows a totally different analysis. And the problem: no matter where I look on the Internet a pink noise spectrum slopes from high to low / left to right (iZotope correct???) and not from low to high / left to right (as is being shown by Voxengo and Fab Filter). I also tested using a pink noise generator (VST). Same result. Any ideas??? There are numerous articles and videos on the Internet that use the exact same plugins (spectrum analysers and pink noise generators) and in those articles and videos the spectrums resemble my iZotope display not my Voxengo or Fab Filter display.

Screenshot_2018-06-30 10-37-46.jpg (465.84 KiB) Viewed 4634 times

Regards,

Dale.

[dpaterson]

Hmmmnnn...

Upon further investigation it would appear that this is not a Mixcraft issue. Tested the same in MAGIX Sound Forge Pro 12. Same results. And to add to the confusion: tested with Blue Cat's various spectrum analysis VST's and they all agree with iZotope. Don't quite understand what's going on here. I mean to say: iZotope is supposed to be "Rolls Royce" (which I agree with) but Voxengo and Fab Filter are by no means "below par" either.

Regards,

Dale.

[dpaterson]

Here's the Blue Cat analysis (that agrees with iZotope).

And to be safe: one track is the downloaded pink noise audio track and the other is the generated pink noise (VST).

bluecat.jpg (571.44 KiB) Viewed 4628 times

Regards,

Dale.

[dpaterson]

The plot thickens!!!

Voxengo's SPAN Plus has an option in settings to change the slope of the display and it's default is NOT zero. Check out the video: https://youtu.be/XeFr_Jt4lrI. I'm assuming that there's a good reason for this setting / option (because it's confused me something chronic and had I not found this it would make the plugin useless in my opinion i.e. your mix could appear as having all of the lower frequencies attenuated and the upper frequencies boosted when this is indeed not the case).

I cannot comment on Fab Filter as yet but have contacted both Fab Filter and Voxengo for support on this.

Regards,

Dale.

[dpaterson]

Yeh. Alright. Having great fun and answering most of my own questions.

From the Voxengo manual:

Note that by default Voxengo plug-ins use 4.5 dB per octave slope for the spectrum
display which makes it look considerably “elevated” towards the higher frequencies in
comparison to most other spectrum analyzers available on the market. This setting
can be changed in the “Spectrum Mode Editor” window.

From the Fab Filter manual:

The Tilt setting tilts the measured spectrum around 1 kHz with a specified slope, expressed in dB per
octave. The default setting of 4.5 dB/oct results in a natural looking spectrum, resembling best how
loudness is perceived by the human ear.

None of the above was mentioned in any online tutorial though!!!

Oh well. Problem(s) solved. Will leave the posts here though (you never know: somebody else may end up being confused by this at some point).

Regards,

Dale.

[dpaterson]

For the sake of completeness:

"One interesting point to note is that a mix that is EQed to sound flat will not necessarily look so on the graph. This is because our hearing does not respond to all frequencies equally; for instance, 80dB at 100Hz sounds quieter to us than 80dB at 1kHz. In fact, many analysers artificially adjust the slope of the display so that what looks flat on the graph does sound relatively flat to us, and the common setting here is between 3 and 4.5dB/octave, approximating our ears' response. A well-balanced mix viewed using a 3dB/octave slope setting would have a mostly flat appearance, sloping neither up nor down by very much."

Hence the slope settings (ironically in all BUT iZotope). Then again: were it not for iZotope I'd never have noticed this nor learned about it.

Onto the next phase (no pun intended).

Regards,

Dale.

[Mark Bliss]

Interesting conversation with Dale here Dale....

Actually, quite useful on several points.
On the slope calibration thing, I knew Span did that by default, I was unaware that Pro-Q also did. (I have used it, not extensively enough to become "familiar")
But it simply intends that the graph would look more flat than sloped in actual use. Good thing to be aware of if you are attempting to emulate the sloped graph target huh?

Personal experience, I also messed around with the referencing to pink noise thing for a while (but didnt inhale) and have observed others do the same. I found that most obsessed over it a bit too much and overcompensated....
Taking it too literally and chasing our tails as a result can consume a lot of time cant it.

Its a good idea to be aware of the overall freq trend IMO, and to be able to corrolate that idea with the natural hearing response (Fletcher-Munson curve) concept. Trouble is this is more of a generalization than rule IMO, and... the whole hearing response things changes with volume anyway.

My suggestion, as I think it helps "shorten the route"
Instead of referencing pink noise, reference some commercial music. Using the meter of your choice (and reference slope, just be consistent I reckon)
I like to use Span, in avg./max mode and study the results. And again, dont chase every little dip and spike, but note the overall trend of multiple examples. A spike at a specific freq could only mean the bassoon wasn't playing that note in the other references.

Then when you look at the results on your own songs, you can notice, "hey- theres too much bass content here" etc. In other words: Focus on the general trends.

[rrichard63]

To add to the confusion, there's "white noise" as well as "pink noise". In a spectrum analyzer (e.g. Voxengo Span), white noise will look flat with the slope set to 0 dB per octave. Pink noise will look flat with the slope set to 3 dB per octave. For some hints as to why, see https://www.acousticfields.com/white-no ... ink-noise/.

[dpaterson]

Hello (again).

Interesting conversation with Dale here Dale....

Hope you enjoyed it!!! LOL!!!

Good thing to be aware of if you are attempting to emulate the sloped graph target huh?

My point exactly!!! I suppose the lessons here are a) don't take the word of at at least the first five YouTube tutorial videos made by the next budding sound engineer / expert and b) read the f*****g manual ("RTFM"). I must say: the chap at Voxengo came back to me within an hour to explain why it's done this way i.e. why there is a slope option. I'm figuring the chaps at Fab Filter just think I'm an idiot!!! LOL!!! Am VERY surprised that my plugin "supplier of choice" (iZotope) don't have this "feature" though.

From what I gather though: using this pink noise balancing thingy can save quite a bit of initial time. Only thing that's confusing me about it (and somebody actually did ask this in the comments section under one of the YouTube videos on this subject but no answer was given): SURELY if you bring up each track to just below the pink noise (so that you can no longer hear the track) then this should be done AFTER you've applied effects to the track (for the simple reason that an effect may push the level to ABOVE the pink noise)??? So far as I can tell: all the so-called tutorials on the subject say to do this pink noise adjustment BEFORE applying any effects. Seems counter intuitive to me (but I'm guessing that one of you will put the record straight!!! LOL!!!).

And yeh (rrichard63): got myself a pink noise and white noise generator (for use as per one of my other threads re: using the "shavering technique" for mic. placement in front of my amps. I was thinking that maybe white noise would be the perfect signal to be using) (not tested though as yet) (although I have to say that thus far: my ears have done a darn good job of letting me know where the "sweet spot" is on my amps. before employing the "shavering technique" i.e. the technique puts the mics. in the exact same places as I had them anyway!!! LOL!!!).

Regards,

Dale.

[Ian Craig]

Yes, Melda's MSpectralDynamics has a default slope setting of 3dB in the Spectrum area of it's edit page. It's worth checking out the demo as they have lots of free plugins and the paid ones keep working with only occasional added noise (also they have more options than anything else in the world, so are handy for learning).
iZotope are gougers really. I say that having bought the Advanced versions of RX6 (which due to a bug could destroy all of anyone's work over years if they believed they could batch process wav files to mp3 and then delete them as it produced 32kbit mp3 files - this on the advanced £1100 version), Ozone 8, Iris 2 and stupidly the rather pointless Neutron 2. Also Vocalsynth only a month or so before they were looking for £170 to upgrade to Vocalsynth 2. I crazily bought the Creative Suite upgrade to get it (which included Iris 2 that I paid £125 for last year), so if they don't respond to your emails it's probably because they haven't returned from their round the world cruise at my expense)

[dpaterson]

Hi Ian.

Sorry but had to smile at your iZotope comments!!! LOL!!! Oddly enough: I read about your conversion issue on another thread last night. Still. I like their stuff (but only really use Ozone 8 Advanced and Insight). And yeh: I don't quite get the reason for Neutron (was fun to play with when I first got it but that's about it). And yeh: between yours and my "investment" in iZotope hey should be able to spend a little more time on their cruise!!!

Indeed I've been looking at Melda's stuff (they have an automatic phase correction plugin and also their pink noise and white noise generator). Is there stuff any good??? Only reason I ask is because their installation application and method of trying to sell the plugins looks "tacky" to me and the plugins themselves don't seem "polished" (in appearance) when compared to the likes of iZotope (sorry!!!) and Fab Filter (but maybe it's just me i.e. I'm a real sucker for stuff that LOOKS nice).

Regards,

Dale.

[Ian Craig]

It has the most configurable GUI there is and there are endless windows of controls deep in the better ones, enabling more configurable sound processing than I've been able to get my head round in the last 2 years. They have built in limiters in all of them I think (which has to be turned on, as it's off by default, I don't know why), Left/Right, Mid/Side, Left+Right-, Mid, Side, Mid (Zero Side), Side (Zero Mid) selectable option via a single setting on the side, A-H comparison settings, multiple Modulation options the time scale of which (up to 128 beats I think) can be multiplied by up to 64x. I think all this speaks for itself. I don't notice any quality deficit in comparison to anybody else's stuff, their convolution reverb is the best I have come across for producing ambient tunnel like effects and mostly their plugins are very stable. Just run the installer and select all the free plugins and/or any others of interest. It's all in one installer, so you only need to re-download it when it gets updated.

[dpaterson]

Thanks Ian. Much appreciated.

Regards,

Dale.

[Mark Bliss]

Hmm. We wandered here. Surprise!

Getting back to the mixing with pink noise effort, I would reiterate- it may be a useful exercise for learning some things, but I couldn't emphasis more the suggestion to not get too hung up on it. Its a side path at best and not a "trick" I would suggest using regularly.

And even as far as using a spectrum analyzer at all goes, keep in mind, two very different sounds can look very similar on the meter. Just as a sound that is peaking at -4 can be processed in a way that makes it peak far lower and yet sound louder..... Important concepts to grasp IMO.


What I am getting at overall is that this is an EXTREMELY multifaceted craft, and any thought that any single tip, trick or method is a "silver bullet" solution is going to send you down a helluva rabbit trail.

[dpaterson]

Hmmmnnn...

I thought it only fair to retract my observation re: Melda's plugins.

They're good. For sure. Very good. Been working much with their phase alignment plugins this weekend. The words "as if by magic" come to mind. Took two sound files, offset them, and loaded their "MAutoAlign" plugin. With only a little bit of experimentation: it corrected the phase issue 100%. Only problem: their stuff ain't cheap not by any stretch on the imagination (not with our exchange rate anyway).

Anyways. Like I said: only fair.

Regards,

Dale.