How to use a spectrum analyzer for mixing without mixing with your eyes

A spectrum analyzer shows how much signal energy exists at different frequencies. Use it to investigate a question you already hear—such as “what blooms on this bass note?” or “where do the vocal and guitar overlap?”— then make the decision by listening in context.

It is not a score for mix quality. A curve can look smooth while the music feels lifeless, or look irregular while the arrangement sounds exactly right.

What the graph actually means

Most real-time spectrum analyzers display:

• frequency from low to high on the horizontal axis;
• level or magnitude on the vertical axis;
• a changing line or filled area representing the current measurement.

The display is usually calculated with a Fast Fourier Transform. It divides a short window of audio into frequency bins. That window creates a tradeoff: more frequency detail requires more samples, which reduces time precision and slows the visible response.

Set the analyzer before reading it

The same audio can look different under different display settings.

Resolution or FFT size

Higher resolution separates nearby low-frequency components more clearly, but updates more slowly. Lower resolution follows transients faster but groups more frequency information together.

Use high resolution for a sustained bass note or low-end resonance. Use a faster, lower-resolution display when examining the timing of drums or brief events.

Speed, averaging and smoothing

A fast release shows movement and transients. A slower display or longer average makes overall tonal shape easier to inspect. Peak hold or freeze can capture a brief event, but a held maximum is not the same thing as average balance.

Tilt

Many analyzers tilt the display so music appears more visually balanced. FabFilter, for example, documents a default 4.5 dB-per-octave tilt intended to look closer to perceived loudness. Compare curves only when their tilt and other settings match.

Range

A narrow vertical range exaggerates small differences. A wide range can hide them. Keep the range consistent during comparisons.

Five useful mixing jobs

1. Locate a changing resonance

Loop the problem section, listen first, then watch for a peak that corresponds to the unpleasant moment. Freeze can help capture it. Confirm with a narrow band audition at a safe level rather than cutting every visible spike.

Natural instruments create harmonics; peaks are not automatically defects.

2. Understand low-end balance

Compare kick and bass individually and together. Look at where their fundamentals and harmonics fall, but also inspect their envelopes. Two sounds can occupy similar frequencies without conflict when they occur at different times.

A higher FFT resolution can make low-frequency components easier to distinguish. It does not tell you which instrument should be louder.

3. Investigate masking

Overlaying spectra can reveal regions where two sources have strong energy. Treat that as a listening cue, not proof. Temporarily turn one source down. If the other becomes clearer, decide whether arrangement, level, panning, EQ or dynamic ducking is the least damaging fix.

4. Compare before and after processing

Use pre/post views with the same scale and settings. Level-match the processed and unprocessed signal. Louder playback changes perception and can make a processor seem more effective even when the tonal move is not helping.

5. Study a reference

Choose a reference with a related genre, arrangement and loudness. Match playback level and analyzer settings. Compare broad tendencies across several sections rather than copying one frozen curve.

A mastered reference and an unfinished mix are at different production stages. Do not force the mix into the reference shape with dozens of narrow EQ bands.

Spectrum analyzer vs spectrogram

A conventional spectrum emphasizes frequency and level for the current moment or average. A spectrogram adds time as an axis and often represents level with color. Spectrograms are especially useful for seeing intermittent noise, clicks, hum lines, harmonics or frequency events that evolve across a phrase.

Use the view that fits the problem. A real-time curve is quick for tonal balance; a spectrogram is clearer when timing and history matter.

Mistakes to avoid

• Chasing every peak. Harmonics and musical notes should create peaks.
• Trying to make the curve flat. A flat visual spectrum is not a universal sonic target.
• Comparing unmatched settings. Tilt, resolution, speed and range change the display.
• Ignoring loudness. Match playback level before comparing tonal balance.
• EQing in solo only. A strange-looking solo sound may fit the arrangement.
• Treating overlap as failure. Shared frequencies are normal; audible masking is the actual problem.

A practical two-minute workflow

1. Describe the audible problem.
2. Choose analyzer settings appropriate to its frequency and duration.
3. Loop a section that contains the problem and a normal section.
4. Compare pre/post or source/reference at matched level.
5. Make one small change.
6. Close or ignore the analyzer and listen.
7. Keep the change only if the mix improves.

SPC SpectraScope provides spectrum, scopes, heatmaps, stereo and phase views, loudness and true-peak metering in a one-to-four-panel workspace. Its audio path is designed as bit-transparent, zero-latency pass-through, so the analyzer can inspect without processing the signal.

If the analyzer reveals low-level noise between vocal phrases, continue with noise gate vs denoise before choosing a cleanup tool.