Frequency spectrum analysis, explained for producers

Frequency spectrum analysis breaks your mix into bands — sub, bass, low-mid, mid, high-mid, air — and shows how energy is distributed across them. A "pro" spectrum isn't flat; it follows a natural downward tilt, and the goal is a balance that matches how your genre is actually mixed. Unveil measures your spectrum, fits its own slope, and flags the bands that genuinely deviate — not the normal tilt.

Why frequency balance comes first

Tonal balance is the distribution of energy across the frequency range, and it is the single biggest determinant of whether a mix feels right on different speakers. A track with excess low-mid energy sounds boxy on every system. One with too little sub sounds thin on big speakers and fine on laptops, which is a misleading result. A spectrum analysis shows where energy accumulates so you can correct it with intent instead of guesswork.

Measurement does not replace listening, but it catches problems the room and your ear adaptation hide. After an hour of mixing, your hearing has acclimated to whatever is wrong. A spectrum is the same on hour six as on hour one.

The six bands and what lives in each

Sub, 20 to 80 Hz: the weight of kick drums and 808s, synth bass fundamentals, and the lowest organ and bass notes. You feel this range more than you hear it. Most small speakers cannot reproduce it.
Bass, 80 to 250 Hz: bass guitar body, kick punch, the warmth of male vocals and the low end of guitars and piano. Too much here makes a mix muddy and heavy; too little makes it thin.
Low-mid, 250 to 800 Hz: the body of nearly every instrument. This is where mud and boxiness accumulate, typically around 250 to 500 Hz, because many instruments overlap.
Mid, 800 to 2500 Hz: the core of vocals, snare, guitar and keyboard definition. Excess sounds nasal or honky; a deficit makes the mix feel distant.
Hi-mid, 2500 to 6000 Hz: presence, attack, consonants and vocal clarity. The ear is most sensitive here, so small boosts have large effects and excess becomes harsh and fatiguing.
Air, 6000 to 16000 Hz: cymbal shimmer, breath, sibilance at the lower end, and sparkle. Excess sounds brittle or hissy; a deficit sounds dull or closed.

What a healthy balance looks like

Natural sounds and most finished music have more energy at low frequencies than at high ones. On a spectrum analyzer with logarithmic frequency spacing, a typical commercial mix slopes downward from the bass region toward the highs. A slope in the neighbourhood of -3 to -4.5 dB per octave is a common description of music averages. Pink noise, which falls at 3 dB per octave when plotted this way, is a convenient reference because it sounds perceptually even.

This tilt varies by genre. Dark, bass-heavy hip-hop and trap carry more sub energy and less air. Acoustic and classical recordings are often flatter in the highs. Bright pop leans toward more hi-mid and air. Treat any slope as a range to compare with references in your genre, not as a target to hit exactly.

Spectrum tilt is the overall steepness of that slope. A tilt that is too steep means a dark, heavy mix; too shallow means thin and bright. Checking tilt before detailed EQ keeps you from making twenty small moves that really needed one broad shelf.

Reading a real-time analyzer versus a long-term average

A real-time analyzer (RTA) shows the spectrum as it is now. It is useful for spotting a resonance on a single sustained note, a harsh hi-hat, or a feedback tone. It jumps with every transient, so a snapshot of a kick can make the whole mix look bass-heavy.

A long-term average smooths the spectrum over many seconds or the whole track. It reveals the tonal character of the song, independent of individual notes, and is the right tool for comparing against a reference. Use an average of at least a full chorus, and ideally the whole song.

Resolution settings matter. A narrow FFT window shows fine detail at low frequencies but updates slowly; a short one is quick but blurs the bass. Fractional-octave smoothing (one-third octave is common) makes the display closer to what hearing perceives and is easier to read than raw FFT bins. Smoothing also hides narrow peaks, so switch resolution when hunting a specific resonance.

Masking: why the spectrum fills up

Masking happens when two sounds share a frequency region and the louder one makes the quieter one harder to hear. A vocal and a rhythm guitar both sitting around 1 to 3 kHz will fight, and the guitar often wins when you want the vocal. The spectrum shows a peak in that region but does not tell you which instrument is responsible, so solo or mute elements to find out.

The fix is usually subtractive and complementary. Cut the competing instrument a few dB where the lead lives, rather than boosting the lead and adding more energy to an already crowded area. Pan or balance can also reduce masking without any EQ at all.

Step-by-step workflow and a worked EQ example

Worked example: a mix has an average low-mid (250 to 800 Hz) level about 4 dB above the reference, while the hi-mid is 2 dB below. Soloing shows a build-up near 350 Hz, from guitars and keys. A 3 dB cut at 350 Hz with a Q of about 1.2 on the instrument bus reduces the boxiness. A 1.5 dB shelf boost above 3 kHz, applied to the vocal rather than the master, restores presence. After re-measuring, the low-mid is 1 dB above the reference and the hi-mid is level, both well within normal variation.

Choose one or two reference tracks in the same genre and match their loudness to yours.
Display a long-term average of the whole song, using one-third octave smoothing.
Compare band by band across the six ranges. Note where yours is more than 3 dB above or below the reference.
Decide whether the difference is a problem or a style choice. A deliberately dark track need not match a bright reference.
Correct broad differences with gentle shelves or low-Q bells before any narrow cuts.
Re-measure, then listen with the reference at matched loudness.

Common mistakes

Treating the reference curve as a target. A reference shows what is plausible, and your song has its own arrangement.
Comparing at different loudness. A louder reference looks and sounds like it has more bass and treble.
Over-reacting to a narrow peak that is part of the instrument's character, such as a bass note.
Boosting to fill a gap instead of asking whether a cut elsewhere would reveal what is already there.
Judging sub energy on headphones or small speakers that do not reproduce it, then boosting blindly.
Equalizing the master to fix problems that belong to a single track.

Checklist

Average spectrum slopes smoothly with no single band standing out by more than about 4 to 6 dB.
Sub is present on systems that can reproduce it, and does not dominate on systems that cannot.
Low-mid is controlled, with no persistent build-up between 250 and 500 Hz.
Hi-mid gives clarity without harshness on a loud playback.
Air is present without hiss or sibilance.
Checked against a reference at matched loudness.

What you get

The six bands and what lives in each — and what "too much 250–500 Hz" actually sounds like.
Why raw band levels mislead: every mix has a downward tilt, so real problems are deviations from your track's own slope.
How masking shows up in the spectrum when two elements pile into the same band.
A free spectrum readout against your genre's reference curve, with the exact bands to cut or lift.

Questions

What should a mix spectrum look like?

Not flat — a gentle downward tilt of roughly 3–4.5 dB per octave is normal. What matters is that no band juts out or dips relative to that slope, and that the overall balance fits your genre.

Is a frequency analyzer enough to mix?

It tells you what is unbalanced, not why. Unveil pairs the spectrum with genre targets and time-localized conflict detection so you know which element to fix and when.

Analyze your mix — free