Mixing & Mastering Series
Using Spectrum Analyzer Slopes
Concepts, Tools & Techniques
Spectrum analyzers are a must-have plugin tool in your mix environment, giving you a visual channel of commentary on what you think your ears are hearing. Voxengo’s SPAN is a popular free one, and Waves’ PAZ Analyzer is a popular not-free one that is often on sale. There are many other free ones, like Oscarizor and MAnalyzer but if you do acquire a large set of analyzers, load one plugin at a time into a project to make sure they’re not the root cause of Free VST DAW Crash Syndrome.
Most plugin manuals don’t delve into the general mix context, since the goal is usually just to explain what the components of the UI are. How or why to use the UI elements is left your general knowledge or ability to find the answers you need in a web search. If you insert SPAN onto a track in your mix, for example, you won’t know just by looking at its main UI that its slope is set to 4.5 dB per octave, because that’s a hidden setting.
This matters because you may have read in the past that to achieve tonal balance in a mix, you want a frequency spectrum similar to pink noise (equal energy per octave). But you might also have read that you should have flat or relatively equal frequencies across your mix spectra, which visually resembles white noise (equal energy per frequency). In short, if you don’t understand the slope, you can end up precisely creating an unbalanced mix in your efforts to create a balanced one using a spectrograph!
Many spectrum tools default to 3dB per octave, as that renders a plot more associated with classic pop sounds, while many of today’s lush music productions reputedly give a flatter frequency visualization with the spectral analyzer slope set to 4.5 dB per octave. If you’re using a spectrograph to guide your mixing decisions, in part you want to decide whether you want your mix to reference a more traditional pop vibe or tilt in a more contemporary flavor rich in very high frequency spectra, which producers seem drawn toward perhaps because of the higher sampling rates available today (e.g 192 kHz) or certain genre inclinations towards edgy high partials often produced by wavetable synthesis (e.g dubstep).
A good place to start is with a pink noise sample loaded into you DAW, which you can download either here or here. Loaded onto some tracks they might look like this:
In many a YouTube video or mixing article, you’ll be instructed to obtain a frequency profile for your mix that slopes gently downward across the spectrum from the lows to highs, in the manner of pink noise. However, that’s ONLY if you have your spectrum analyzer set to a slope of 0. Below, I’ve opened up SPAN’s hidden settings (behind its cog wheel at top right of the main window, not the cog wheel of the plugin wrapper at top left) to show how a slope of 0 dB/octave gives you the plot of pink noise that you may have seen a dozen times before (the slope dial is at the bottom right of the figure, where I’ve opened up the hidden settings).
Mixing to a downward slope can be awkward, since going for tonal balance in a mix is somewhat metaphorically at odds with using an unbalanced (tilted, sloped) visual representation. A better representation, if you are aiming for a mix with tonal characteristics similar to pink noise — which is what you often obtain during moments of peak intensity such as the drop, or what you might see if your representation displays peak frequency values that have played out over the entire duration of a mix with a Peak Hold line, as with the PAZ Analyzer — is to set the slope in the analyzer to 3 dB/octave, which will present pink noise spectra as horizontally flat.
BTW, in these pink noise plots you are also seeing a dip in the very low end produced by a high pass filter, which I apply (with various settings, depending on the track’s aesthetic) to all my mixes to banish the deep chthonic mud tones. The free Hofa System Basic bundle has a great little parametric EQ for this with a very steep roll-off which is causing the low-end dip you are seeing in the pink noise spectra. I love these 48 dB cliff dive low cuts, as they’re excellently extreme.
As mentioned above, SPAN’s default slope is 4.5 dB/octave so if you run the same pink noise through this setting you will get an upward slope.
It should be clear that if you are using a spectrum analyzer to inform your mix decisions, you can get a very different results at these different slope settings if your aim is to achieve tonal balance by getting the frequencies to be evenly distributed horizontally across the spectrum. Your mix will tilt duller or brighter compared to what the articles and YouTube videos said would happen, because they didn’t discuss the slope of the spectrum plot.
If you paid for a plugin like the PAZ Analyzer (note that usually the more you pay for a plugin, the less it tends to cause Free VST DAW Crash Syndrome…) the spectrum plot looks like this:
If you open up PAZ PDF manual, and do a word search on “slope,” no results are found!
It’s worth double checking the manual sometimes to confirm that your eyes are not crazy by not finding the hidden slope setting. The manual says that the PAZ uses wavelets to obtain frequency band level information, which is useful new info. Note that the SPAN uses the Fast Fourier Transform to display its plot (FFTs).
According to a Google search, wavelets are faster to compute than FFTs, which is kind of funny because it’s faster than another process which calls itself ‘fast!’ Maybe wavelet plots don’t use slopes? I confess I don’t know (just because I write blog posts doesn’t mean I can’t admit my ignorance when I stumble across it). Here’s the PAZ manual on the topics of wavelets and pink noise:
Optimal time and frequency resolution in the PAZ is achieve by using wavelet techniques (as opposed to FFTs). This lets each band update independently as fast as possible for its frequency resolution.
A note on Pink Noise measurements: Pink noise is defined as noise that has constant energy inside any frequency band with a constant Q (i.e. frequency/bandwidth=a constant value). Pink noise is an important audio signal since it represents the average spectrum of a typical audio signal, and because the ears analyze the sound in approximately constant-Q bands. Because PAZ has almost constant-Q bands, a pink noise input will show up as a flat spectrum when no weighting is applied.
Voxengo offers no real justification in its manual for the 4.5 dB setting. It just is what it is:
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.
Of course, if you’re using a reference track to guide your mix, it may not matter what your spectra slope setting is if your aim is just to match it. Sloping your mix spectra down, up, or horizontally across in pattern matching to your reference track won’t matter much so long as your reference is going through the same spectral analysis of your mix.
Historically it’d be interesting to know what technologies and music trends are the landmarks for the shift from 3 dB to 4.5 dB in the ‘elevated’ direction of the spectral slope. If you’re only using your ears these slopes don’t matter of course, as they’re just a way of scaling information visually within a window — i.e. the pink noise sounds exactly the same, no mater what the visual slope is doing.
The dangers emerge when you think your frequency balance should look a certain way in the spectral plot and you start to make mix decisions based on an imagined visual and you’ve not taken into account how frequencies are being represented in the plot. If the eyes are telling you the mix looks right but the ears are protesting that something’s wrong, reference your spectrum analyzer to some pink noise and check for hidden slope settings, to reset your ear-eye communication.
Related Articles
Reverse Engineering EDM’s Mix Profile
Knobs, Curves and Spectra | Notes on EQs and UI
Muscle and Subtlety | varieties of parametric equalizers in your mix
Smashing Glue Compressors on the Drum Bus
Breaking the Rules with Room Acoustic Conventions
Let’s Ditch “Master / Slave” Language from Tech, and Replace it with “Boss / Sub.” Sync Sux.
