You don’t learn 200 numbers. You learn about eight landmarks, and you interpolate everything else from there. That’s the whole method, and it’s why engineers can say “that’s around 400” without owning a spectrum analyser.
Nobody has a mental card for 437 Hz. What they have is “higher than the boxy one, lower than the honky one” — and 400 is a good guess for that gap. Get the eight landmarks solid and the in-between answers stop being guesses. They become arithmetic you do by feel.
What are the eight landmarks?
Roughly 60, 120, 250, 500, 1k, 2–3k, 5–8k and 10k+. Notice they’re about an octave apart — each one is double the last. That spacing is the point: an octave is a big enough step that the two sound clearly different, and small enough that everything in between is one halving away from a landmark you already know.
Here’s what each one does, in the plainest words that still mean something. Treat these as rules of thumb, not laws — the edges blur, and different material puts the same problem in a slightly different place.
| Band | What a boost sounds like | What goes missing when you cut it | Where you hear it first |
|---|---|---|---|
| ~60 Hz | Weight. You feel it more than hear it. The floor moves. | Size. The track sounds like it’s coming out of a laptop. | Kick fundamental, 808s, sub bass |
| ~120 Hz | Fat and round. The thump you nod to. | Body. The kick becomes a click with nothing under it. | Bass guitar, kick body, floor tom |
| ~250 Hz | Muddy, boxy, like the mix is in a cupboard. | Warmth first, then clarity arrives. This is the classic cut. | Acoustic guitar body, male vocal chest, whole-mix mud |
| ~500 Hz | Hollow and woody. Someone talking down a tube. | Substance. Things get thin but oddly cleaner. | Snare body, toms, nylon guitar, room mics |
| ~1 kHz | Nasal and plastic. Cheap phone speaker. | Not much, at first. Then the thing sounds scooped and fake. | Vocal nasality, piano mids, snare pop |
| 2–3 kHz | Forward and aggressive. Words jump out. Tiring fast. | Intelligibility. The vocal steps behind the band. | Vocal consonants, guitar pick attack, snare crack |
| 5–8 kHz | Sharp and spitty. Every “s” and “t” gets a point on it. | Edge and detail. Cymbals turn to wash. | Hi-hats, sibilance, cymbal edge |
| 10 kHz+ | Air and sheen. The space around the sound opens up. | Nothing obvious — until you A/B it, and then it’s a blanket. | Cymbal shimmer, vocal breath, reverb tails |
Learn those descriptions as words you’d actually say. “Boxy” and “honky” and “spitty” aren’t jargon for its own sake — they’re handles. When something in a mix bothers you, the fastest route to fixing it is having a word for the annoyance, because the word maps to a number.
How do you actually practise this?
Boost hard, sweep, and name the band out loud before you look at the screen. That’s the drill. Everything else is a variation on it.
Set a bell filter with a narrow Q — somewhere around 4 to 6 — and about +12 dB of gain. Sweep it slowly across a loop you know well. At that setting the boosted band screams; you’re not trying to make it subtle, you’re trying to burn the association between a sound and a number into your head.
When you can name the band before you look, roughly nine times out of ten, make it harder in this order:
- Drop the boost from +12 dB to +6, then to +3.
- Widen the Q from 4–6 down to about 1. Broad moves are much harder to place, because they don’t ring.
- Switch from boosts to cuts.
- Finish with broad, gentle cuts — a 2 dB dip at Q 1. If you can hear those, you’re done.
Do it on varied material. A 3 kHz boost is obvious on a busy synth and almost invisible on a sparse kick-and-hat loop, so if you only ever practise on one loop you’re learning that loop. Render a handful of one-bar sources — drums, a pad, a vocal, a full mix — and rotate them. Our free Serum 2 preset pack is an easy source of sounds you’d actually use in a track, and the Serum 2 bundles cover the sparse-to-dense range if you want more to sweep through.
Why are cuts harder to hear than boosts?
Because a boost adds a new event and a cut only removes one. Your ear is very good at noticing something arriving and much worse at noticing something leaving.
A narrow boost creates what is effectively a resonance — a tone that sits on top of the material and rings. You can hum it. A narrow cut just makes a hole, and the sounds either side of the hole tend to fill in the gap perceptually, so there’s nothing to hum. On dense material a 3 dB dip can be almost inaudible in isolation and still change the mix once everything else is playing.
This matters practically, because most real mixing is cutting. So train boosts to build the map, then spend the bulk of your time on cuts, where the actual work happens.
Why level-matching decides whether the drill works
Because if you skip it, you learn to detect loudness instead of tone, and loudness is the wrong skill. A +12 dB narrow boost makes the whole thing measurably louder, and louder reliably sounds better — fuller at the bottom, brighter at the top, more “there”.
Two habits fix most of it. First, trim the output down when you boost so the A and B are roughly the same subjective level. Second, don’t judge on the first second — that’s when the loudness jump dominates. Let the loop go round twice and listen to what colour changed, not whether it got bigger.
The same rule applies when you’re using EQ for real. If you bypass and the processed version is 1 dB louder, you’re not comparing EQ moves, you’re comparing volumes, and the louder one wins every time.
The trap: naming the band by how loud it got
The commonest way to fool yourself in frequency training is to answer from loudness rather than tone. If a trainer always boosts by the same amount, some bands simply get louder than others — a fixed boost around 2–3 kHz will jump out more than the same boost at 60 Hz on most speakers, and you can score well by learning that pattern rather than the sound.
You’ll know it’s happening when your accuracy collapses the moment the boost gets smaller, or when you switch from headphones to speakers and suddenly you’re guessing. The fix is to force the cue to be timbral:
- Hum or sing the band you think it is, then check. If you can pitch it, you’re hearing tone.
- Vary the boost amount randomly, so loudness stops being a reliable clue.
- Use material with different spectral shapes, not one comfortable loop.
- Answer with a word (“boxy”, “nasal”) before you answer with a number.
Small caveat about small speakers and phones: a lot of them don’t reproduce the bottom landmark at all, so drilling 60 Hz on laptop speakers is training on something you can’t hear. Do the low bands on headphones.
Which tools drill this, and what they cost
Three do this specific drill well and two of them are free. All checked August 2026.
| Tool | Price | Where it runs | What the frequency drill looks like |
|---|---|---|---|
| Monosounds Ear Training (ours, so weigh that) | Free | Browser, phone or desktop | Boosts and cuts as levels across 500 levels; miss one and a hint describes what that range sounds like |
| EQ Trainer | Free | Browser | Boost identification with a bypass A/B, difficulty tiers up to “God mode”, pink noise or music-track modes, daily challenges |
| EarQuiz Frequencies | Free, open source (GPL-3.0) | Windows, macOS, Linux desktop | Golden-Ears-style technical drills, pink noise generator, loads your own WAV/FLAC/MP3 files |
Our own ear training game structures the boost-then-cut progression as levels so you don’t have to decide when to make it harder, and the hint you get after a miss describes the range in words — the “boxy”, “nasal” vocabulary above — rather than just telling you the answer was 500. It’s a game, not a course. Nobody lectures you.
EQ Trainer’s bypass A/B is the closest thing to the manual drill, and EarQuiz is the one to grab if you want it offline and want to load your own files. If you’re comparing the wider field, including the paid options, we went through them in our SoundGym alternatives roundup.
Honestly though: you can do this entire drill with the stock EQ in your DAW and a loop. The tools mainly save you from marking your own homework.
FAQ
How many frequencies do I actually need to memorise?
About eight. Roughly 60, 120, 250, 500, 1k, 2–3k, 5–8k and 10k+, which are spaced about an octave apart. Everything else you estimate by deciding which two landmarks it sits between. Nobody identifies 437 Hz directly — they hear “between boxy and honky” and say 400.
What Q and gain should I use when sweeping?
Start at a narrow Q around 4 to 6 with about +12 dB of boost, because that makes the band ring and gives you something to hum. Once you’re reliably right, drop to +6 and then +3 dB, then widen the Q toward 1. Broad, gentle moves are the hardest to place and the most useful to know.
Why can’t I hear EQ cuts as easily as boosts?
A boost adds a resonance you can hear arrive; a cut just leaves a hole, and neighbouring frequencies mask it. That’s normal and it doesn’t mean your ears are bad. Build the map with boosts first, then spend most of your practice on cuts, since cutting is what you’ll actually do in a mix.
Do I need special headphones for frequency ear training?
No, but you do need something that reproduces the band you’re training. Laptop and phone speakers often produce almost nothing at 60 Hz, so drilling the bottom two landmarks on them is pointless. Any honest headphones work; what matters far more is using the same pair consistently so you learn one reference.
How long before I can name frequencies reliably?
Most people get the eight landmarks recognisable within a few weeks of short daily sessions, and then narrow cuts and broad gentle moves take considerably longer. We’ve written up realistic timelines in how long ear training takes. Ten focused minutes a day beats a two-hour session on Sunday.
Does frequency identification actually make my mixes better?
It makes you faster and more certain, which is not the same as making you tasteful. Knowing the problem is at 250 Hz saves you the sweep, but deciding how much to cut is a taste call you build by finishing tracks. The rest of the picture is in our guide to ear training for mixing, and you can start the frequency levels in our trainer without an account.
I run Monosounds.studio and make the packs myself. Ten-plus years of producing, 97+ releases, 16,000+ downloads. Every Serum 2 preset we ship passes a CPU check on a mid-range laptop and gets used in a real track before release. If it does not survive a mix, it does not go in the pack.
More about Maxim & the studio →
