Audio analyzers · Guide
How to use a spectrum analyzer for mixing
Your ears decide whether a mix works, but they get tired, and your room and speakers lie to them, mostly in the bass. A spectrum analyzer does not get tired. It shows the low end your speakers cannot play, the narrow peak that makes a vocal harsh, and the rumble that eats your headroom. This guide shows which settings to use, the five problems an analyzer finds fastest, and where it can mislead you.
Try it
Open the live analyzer with the Music preset, then feed it your mix. On a computer with Chrome or Edge, pick Tab audio as the source and play your song in another tab or a web player. Or pick Audio file and open a bounce of your mix. Nothing is uploaded.
- Press the button below. The analyzer opens with the Music preset.
- Choose the source: Tab audio, Audio file, or the microphone near your speakers.
- Turn the Tilt knob to 4.5 dB and set Averaging to exponential (or linear over 16 frames or more), as the next section explains.
Best analyzer settings for mixing
A mix analyzer should show tone balance, not every flicker. Start with these settings and change them only when you have a reason.
- FFT size 8,192 at 44.1 or 48 kHz. That gives bins of about 5 to 6 Hz, enough to see the kick and the bass as separate peaks, while the display still follows the music. Go to 16,384 or 32,768 only when you study the lowest octave; the display then reacts more slowly.
- Tilt (slope) 4.5 dB per octave. Music has much more energy in the bass than in the treble, so on a flat display a good mix looks like a hill falling to the right. A 4.5 dB tilt lifts the right side until a typical mix looks roughly level, and then any bump or dip stands out.
- Slow averaging. Exponential averaging over a second or two, or a long-term average over a whole chorus, shows the balance. Fast settings show the notes, which is useful for finding clashes but useless for judging tone.
- 1/6-octave smoothing for balance, off for hunting narrow peaks. Smoothing merges nearby bins so broad trends are easy to read.
- Log frequency scale, 20 Hz to 20 kHz, and a display range of about 60 dB. Anything more than 50 or 60 dB below the loudest part rarely matters in a mix.
Watch two views. The averaged spectrum tells you whether the mix is too dark, too bright or heavy in one region. A spectrogram, or the live spectrum with no averaging, shows which notes collide and when. If you are not sure what lives in a range, keep the EQ frequency chart open next to the analyzer.
A simple routine works well. Mix by ear for a while first, at a moderate level. Then take a short break, play the loudest section with the analyzer open and look at the averaged curve from left to right: below 30 Hz, the low end, the 200 to 400 Hz area, the 2 to 5 kHz area and the top octave. Write down anything that sticks out by more than 3 dB, listen for it on its own, and only then reach for EQ. Checking takes two minutes, and it stops you from chasing problems that are only in your room.
Five problems an analyzer finds fast
Each of these is easy to miss by ear on small speakers or in an untreated room, and easy to see on a screen once you know where to look.
Low-end build-up
Mud lives between about 200 and 400 Hz. Guitars, keys, the body of a snare, the low end of vocals and the room sound of every mic all add a little there, and together they make a mix sound boxy and thick. On a tilted, averaged analyzer it shows as a broad hump of 3 to 6 dB around 250 Hz compared with the rest of the curve, or compared with a reference track.
Do not fix it on the master. Solo the tracks one by one, find the two or three that add most in that range, and cut each by 2 or 3 dB with a wide bell. Small cuts on several tracks sound more natural than one big cut on the whole mix.
Kick and bass clash
The kick drum's thump usually sits between 50 and 80 Hz, and a bass guitar's low E is 41 Hz with strong harmonics at 82 and 123 Hz. When both peak in the same place, the low end gets louder but less clear. Play the kick and the bass together with no averaging and look at the lowest octave: two separate peaks are good; one fat peak that pumps on every kick is a clash.
Decide who owns the very bottom. If the kick peaks at 55 Hz, cut the bass 2 to 3 dB there and let the bass carry 80 to 100 Hz, or the other way round. Sidechain compression, which ducks the bass a little on each kick, solves what EQ cannot.
Harsh peaks
Our ears are most sensitive between about 2 and 5 kHz, so a narrow peak there sounds harsh long before it looks big. Typical sources are a vocal resonance, a ringing cymbal or a bright guitar amp. Switch on peak hold, play the loud chorus and look for a spike that stands 6 dB or more above its neighbours and keeps coming back at the same frequency.
Cut it narrowly, with a Q of 4 to 8 and 3 to 6 dB of cut, on the track that causes it. A dynamic EQ that only cuts when the peak appears is even gentler. Sibilance at 5 to 8 kHz on "s" and "t" sounds is a job for a de-esser rather than static EQ.
Rumble below 30 Hz
Many tracks carry energy far below any note in the song: footsteps through a mic stand, wind, air conditioning, or a synth with a very low tail. Most speakers cannot play it, so you do not hear it, but it still uses headroom and makes a limiter pump. Zoom the analyzer to 10 to 60 Hz. If the level below 30 Hz comes within about 10 dB of the level at 50 to 60 Hz, there is rumble to remove.
Put a high-pass filter at 25 to 30 Hz on tracks that need deep bass, and much higher on tracks that do not: 80 to 100 Hz on most vocals, and often 100 Hz or more on guitars and keys in a full mix.
Missing air
The top octave, roughly 10 to 20 kHz, carries the sparkle of cymbals, the breath in a vocal and the sense of space. Mixes made on dull speakers, or with lots of old analog-style plugins, often fall away steeply up there. On a tilted analyzer a healthy mix stays fairly level up to around 10 kHz, then rolls off gently; a mix with no air drops like a cliff from 8 kHz.
A gentle high shelf of 1 to 3 dB from 10 or 12 kHz usually brings it back. Compare against a lossless reference: many MP3 and streaming files are cut off near 16 kHz on purpose, so a lossy reference makes your mix look brighter than it is.
Matching a reference track
A reference is a finished song in the same style that sounds the way you want yours to sound. Put both through the same analyzer with the same settings, and the differences in tone balance become visible.
First match the loudness. A louder track looks bigger everywhere and also sounds better, which fools both your ears and your eyes. Check loudness in LUFS with a loudness meter and turn the reference down to match your mix. Then compare long averages of the same part, such as the two choruses, not a quiet verse against a loud drop.
Look for broad differences: 3 dB or more across an octave or wider. Those are real tone choices worth acting on. Ignore small wiggles; no two songs share the same notes, so narrow peaks will never line up. To lay the two curves over each other, use compare with a reference, or save one curve as a reference trace in the live analyzer and play the other.
What an analyzer cannot tell you
An analyzer measures energy per frequency, averaged over time. That leaves out a lot of what makes a mix good. It cannot tell you whether the vocal is easy to understand over the guitars, because masking depends on timing and on how loud each part is from moment to moment. It cannot show depth, punch or groove. Two mixes with the same curve can feel completely different.
It also hides the stereo picture: a single spectrum adds left and right together, so a wide chorus and a narrow one look the same. For that, use a phase or correlation meter. And it cannot tell you whether a choice is right for the song. Use the screen to check what you hear, never to overrule it.
Monitors and room first
The analyzer can only check what you feed it. The bigger problem in most home studios is the listening chain. A small, bare room boosts some bass notes by 10 dB or more and nearly cancels others, so you add bass where the room hides it and cut it where the room booms. Check your room with pink noise and an RTA at the listening spot: dips and peaks of 10 dB or more below 200 Hz are the room, not the speakers.
Decent nearfield monitors help more than any plugin. A pair like the Yamaha HS5 has a 5-inch woofer, is rated down to about 54 Hz and is known for an honest, slightly unflattering mid-range, which makes harsh peaks easy to hear. Set the speakers and your head in an equal triangle, tweeters at ear height, and keep them off the back wall. Then the analyzer and your ears start telling the same story. For more tools, see the audio analyzer tools.
We may earn a commission if you buy through these links. It does not change the price.
FAQ
Should a mix look flat on an analyzer?
No. Most good mixes fall gently from bass to treble, about −3 to −5 dB per octave on a flat analyzer.
What slope setting should I use?
A 4.5 dB/octave tilt makes typical mixes look roughly flat, which makes problems easier to spot.
Can I mix with an analyzer only?
No. Use it to check what you hear and to catch things your room or speakers hide.
Which analyzer view is best for mixing?
An averaged FFT spectrum for tone balance and a spectrogram for clashing notes over time.