Frequency Blindfold Test
This frequency blindfold test is the classic “golden ears” engineer drill. We loop a generated broadband source — pink noise, white noise, or a harmonically-rich synth tone — and quietly boost or cut one band with a peaking EQ filter. A/B it against the flat reference, then name the band. Choose your level: the bands narrow from octave to one-third-octave and the gain shrinks from ±12 dB down to ±3 dB as you step up the difficulty.
ℹ This is an uncalibrated practice aid that trains relative spectral discrimination — how well you spot which band changed. The affected band is known exactly because this tool applied the filter; there is no microphone and nothing is recorded. What you hear depends heavily on your headphones/speakers and room, so cheap earbuds will roll off the extremes. EQ words like “warm” or “airy” are conventional, subjective shorthand, not measurements. Your score is a personal practice metric stored only in this browser — not a certification. Use a moderate volume.
Press “New round” to start. One band will be boosted or cut.
Choose your settings, then press “New round” to begin.
Your running totals and best streak are stored only in this browser (localStorage). Clearing site data or using private mode erases them; nothing is uploaded.
How It Works
Every round starts with a generated broadband signal — pink noise (equal energy per octave), white noise (equal energy per Hz), or a harmonically-rich synth tone built from a fundamental plus its integer harmonics. Nothing is sampled or recorded; the source is synthesised in your browser, so there is no copyrighted music involved. That signal is routed through a single peaking biquad filter — the same shape as one band of a parametric equaliser — centred on one randomly chosen frequency, with either a positive (boost) or negative (cut) gain. This parametric EQ approach mirrors the real-world technique used in professional audio engineer ear training courses, where engineers sweep a narrow band of a noise source to develop critical listening skills, frequency discrimination, and frequency recognition by ear.
The A/B buttons crossfade between the flat reference (filter gain = 0 dB) and the processed signal (filter gain set to the round’s boost or cut), with a short click-free ramp so the only difference you hear is the spectral colour. Your job is to listen for the band that gets louder or quieter and identify the frequency band by picking its centre frequency — the core skill of EQ band training. On Easy and Medium, choices come from the nine octave centres (63 Hz–16 kHz) at ±12 and ±8 dB; on Hard and Expert they come from the full set of ISO one-third-octave centres (50 Hz–16 kHz) at the harder ±5 and ±3 dB, with a tighter filter Q to match the narrower band. After you answer, the tool reveals the exact centre frequency, gain and filter shape, and (if you missed) how many cents or octaves your guess was off. To see the spectral shape of each boost or cut visually while you train, you can open the audio spectrum analyzer alongside and route the same source through it.
Two honest cautions. First, this trains relative discrimination, not absolute calibration: because the tool itself applies the filter, the “answer” is exact, but the loudness and audibility of each band at your ears depend on your headphones or speakers, your room, and your own hearing — a 12.5 kHz cut is invisible if your gear (or your ears) already rolls off up there. For best results, practise on closed-back headphones with a flat or neutral response; consumer earbuds with exaggerated bass or treble can fool your perception. Second, the named character of each band (“mud”, “presence”, “air”) is conventional studio shorthand and subjective; the frequencies are exact, the adjectives are not. Treat your score as a personal practice metric, not a measurement or certification. If you want to reinforce the tonal vocabulary alongside this drill, the EQ frequency band trainer walks through the character of each band step by step.