Reading a Spectrum Analyzer When You Only Have Ten Minutes

I can usually tell how a mix was prepped by watching someone use an analyzer. If they stare at the curve, chase every wobble, and nudge until the line is flat, the result almost always sounds stiff. Ten minutes with a spectrum analyzer is plenty, but only if you point it at the right questions.

This is not about pulling up SPAN and matching references pixel by pixel. It’s a radio habit: a quick, repeatable check before a track goes out. Four looks, maybe five. Then you close it and listen.

Set the slopes first, or the whole picture lies to you

Before you read anything, check your analyzer’s settings. FFT size and slope are the two dials that matter most, and the wrong ones will send you chasing problems that aren’t there.

The common sliding “pink noise” tilt (around 3 dB per octave, or 4.5 dB per octave depending on the analyzer) exists so that music reads roughly flat. If you’re on a raw flat FFT, everything looks tilted up toward the highs, and you’ll start cutting treble that’s actually fine. Pick one slope and stick to it. Switching between views is how you convince yourself a mixer is dark when it’s neutral.

FFT size is a resolution/speed trade. Bigger windows give you finer detail but smear fast transients; smaller windows react quicker but blur low-frequency detail. For a quick ten-minute pass, a moderate window (say 2048 to 4096) is enough to see broad shapes. You don’t need surgical resolution. You need to know where the energy sits.

Look one: the overall tilt

Play a section with the most going on: the chorus, the drop, the bridge. Ignore individual peaks for a second. Just look at the falloff from low to high. Does the energy drop off in a roughly even slope, or does it fall off a cliff, or spike somewhere it shouldn’t?

That macro shape tells you more about balance than any single frequency. A mix that slopes smoothly downward reads as full without being boomy. A mix that’s flat across the low mids and then crashes into a dull top is going to sound closed and congested on a radio chain.

If the tilt is wrong, that’s a mastering decision, not a mix move. Broad, gentle shelves. That’s the point: at this stage you’re confirming a shape, not fixing a problem.

Look two: the low end, but only the broad strokes

This is where the ten minutes gets used up fastest. Sub and bass are where an analyzer helps most, and also where people overshoot.

Zoom into the lower register and watch the broad weight around the sub and fundamental zone. You’re not hunting for a single bad frequency. You’re asking whether the weight is roughly where it should be, not pushed too hot in the rumble and not light overall.

Watch the monitor path here, not the analyzer. A reference like the Sennheiser HD 480 Pro, for instance, pushes a little extra weight down at the very bottom but pulls back around 70 to 80 Hz and dips more audibly near 200 Hz. If you’re judging low-end balance on headphones with that kind of response, the analyzer and your ears will disagree. Trust the analyzer for the broad shape and trust your monitors (or a well-measured set) for the judgment. Otherwise you’ll “fix” a 200 Hz dip your cans put there themselves.

One practical rule: if you’re spending more than a couple of minutes on the low end, you’ve stopped looking and started mixing. Stop.

Look three: the upper mids and the presence range

Sweep your eyes up to the presence range. This is where harshness, honk, and vocal intelligibility live, and it’s the part of the analyzer most likely to make you do something dumb.

A little extra around the upper mids isn’t automatically a problem. Many tracks are cut to sound forward there because they need to survive a car speaker or a phone. What you’re checking for is a spike that sounds aggressive rather than musical, the kind that makes you flinch on a long listen.

If something pokes out, a narrow cut is tempting and usually wrong for radio. Broad, small moves. And if it’s only happening occasionally, a de-esser or a dynamic approach beats a static EQ cut, because you’re only taming the offending moment instead of hollowing out the whole track.

Look four: the top, and what the broadcast chain will do to it

Highs are where the analyzer earns its keep, because the broadcast chain is unkind up there. Streaming and radio processing will squash peaks and expose sibilance. So you look for two things: a top end that isn’t unnaturally shelved down (which reads dull once processing bites), and sibilant spikes that will jump out after compression.

Headphone response quirks matter again here. That dip around 5 kHz and peak near 8 kHz on a set like the HD 480 Pro means a perfectly reasonable top end can look wrong on your ears. The analyzer is a check on the monitors, not the other way around. Match them, don’t obey one.

If you use a clipper or saturation to tame peaks, look at how it’s shaping the top. A tool like Shatter Clip splits the signal at 120 Hz, adds tube-style saturation to the sub-bass while keeping high transients cleaner, and shows a 60 FPS oscilloscope with the clipped material highlighted. That visual is genuinely useful: you can see how much peak you’re shaving in seconds, and the auto-gain keeps you from being fooled by level changes. Ten-minute workflow friendly.

What you don’t do with ten minutes

You don’t reference-match. Two analyzers side by side, one on your track and one on a commercial release, is a good way to convince yourself your vision is wrong. Use a reference if you want, but glance at the broad tilt only. Matching curves never sounded like matching records.

You don’t chase single frequencies. If you catch yourself soloing a band or sweeping a narrow Q, close the window and listen. The frequency curve is not the music. Hardware people know this in their hands: you stop looking at the curve and start listening to the sound. The same discipline applies on screen.

You don’t leave it open. The last and most important step of the ten minutes is closing the analyzer and doing one full listen. If the balance feels right, it is right, regardless of what the line looked like.

The short version

Set your slope, glance at overall tilt, check broad low-end weight, check upper-mid presence, check the top against how the broadcast chain will treat it. Broad shapes, small moves, then close it.

Ten minutes is enough to catch the things that will actually bother a listener over a two-hour set. Everything past that is overthinking, and overthinking is what makes a mix sound like it was built by an analyzer instead of a person.

If you want one repeatable habit out of this: before every export, do the four looks, then ask whether anything you saw would survive a listener noticing it. Most of the time the answer is no, and you’re done in three minutes, not ten.