Hearing Range Test

Read this first. This widget is a curiosity tool, not a clinical audiogram. A real hearing test uses calibrated equipment in a quiet booth and measures a threshold level in decibels at each frequency. This one plays an uncalibrated tone through whatever hardware you happen to have, so it measures your headphones, your volume knob and your room at least as much as your ears. It cannot detect, rule out or diagnose any hearing condition. Please do not treat a missing tone as a medical finding — take that question to a professional.

Equipment caveat

• Use wired over-ear or in-ear headphones if you can. Bluetooth codecs often roll off or mangle content above ~15 kHz.
• Laptop, phone and tablet speakers typically produce nothing usable below ~100 Hz or above ~15 kHz. The bottom and top of the ladder will simply be silent on them.
• Background noise masks quiet tones, especially in the bass. A fan or traffic will cost you the low end.
• No equal-loudness compensation is applied. At one fixed digital level, tones near 2–4 kHz sound much louder than tones at 20 Hz or 18 kHz. That is normal ear behaviour, not a fault.

Safe listening

• Start at the default level (18 %) with your system volume low, and raise the system volume only until the 1 kHz tone is comfortably audible. Then stop.
• Do not raise the level to force a high tone into existence. If 18 kHz is silent at a comfortable level, the honest answer is “I don't hear it”.
• Take the headphones off between runs. Tone bursts are fatiguing.

One ear at a time

• Both / Left only / Right only routes the tone to one output channel and leaves the other channel genuinely empty — not quieter, not panned, simply not connected. On headphones that means one ear hears everything and the other hears nothing at all.
• Each ear keeps its own twenty answers, its own fine-search result and its own curve, so you can run the ladder left, run it right, and lay the two on top of each other.
• This only works on stereo headphones. On a laptop speaker pair, on a mono speaker, or on anything that downmixes to mono, both channels reach both ears and the comparison is meaningless.
• Two ears rarely land on exactly the same step, and a difference of a step or two is ordinary. A channel imbalance in your headphones, one earcup sitting differently, or a bit of wax will all produce the same picture. A gap between your ears here is not a finding and not a diagnosis — if it bothers you, ask an audiologist.

A–B two-interval mode

• Instead of asking “can you hear it?”, this plays two intervals in a row, marked 1 and 2 on screen. Exactly one of them contains the tone; the other is real silence. You say which one it was.
• You cannot pass by being optimistic, because you are not judging whether a sound exists — you are choosing between two options, and guessing is right half the time by definition.
• Each frequency runs three trials. All three correct records the step as heard; anything less records it as not heard. Pure guessing gets all three about one time in eight.
• Interval length follows your burst-length setting but is capped at 2 s each, so a trial never drags.
• A–B is slower than yes/no and far more trustworthy. If your yes/no ceiling and your A–B ceiling disagree, believe the A–B one.
• Catch trials are not used in A–B mode, because forced choice already does that job.

Catch trials — the honesty check

• In yes/no mode, roughly one trial in six plays nothing at all. No oscillator is created, so it is silence in the strictest sense.
• If you answer “I hear it” on a silent trial, you are told immediately, the answer is discarded, and the real tone is played so you can answer properly. Nothing is scored against you.
• The running tally sits under the curve. False alarms do not mean you were dishonest — near your limit, the ear and the imagination genuinely blur, which is exactly why the check is there.
• If you collect false alarms, treat the very top of your range as soft. Switch to A–B mode and run it again.
• You can turn catch trials off, but a self-reported hearing test without them is not worth much.

Reading the curve

• Frequency runs across the bottom, one column per ladder step, evenly spaced — the same convention a real audiogram uses.
• A marker on the upper row means heard, a marker on the lower row means not heard. Height carries the answer, so it does not rely on colour.
• Left is a filled circle on a solid line, right is an open square on a dashed line, both-ears is a diamond on a dotted line.
• The shaded block is the neutral reference band, roughly 10–19 kHz, covering where commonly quoted adult ceilings tend to fall. It is context, not a pass mark, and no part of the chart is a clinical audiogram.
• A fine-search result appears as a thin vertical marker in that ear's colour.

Free play and reference tones

• The free-play slider is logarithmic across the whole 20 Hz–20 kHz range. Hold the button to sound the tone and keep holding while you drag — the pitch glides smoothly instead of restarting, so you can creep up on the exact point where it vanishes.
• Free play routes through your ear selection too, and records nothing. It is there to play with.
• The reference row plays everyday landmarks at their exact headline frequency and parks the slider there. Those labels are the dominant frequency only — a real cymbal or mosquito is a broad, noisy spectrum.

Mouse

• Ladder cell — click any frequency to make it the current step and play its tone. All 20 steps stay visible; nothing scrolls.
• Play / Replay tone — plays the current step again as a short burst.
• I hear it / I don't hear it — records your answer, fills the cell in, and auto-advances to the next untested step upward, playing it straight away.
• Start at 20 Hz / Start at 8 kHz — jump the flow to the bottom of the ladder or to 8 kHz, which is where the interesting part begins.
• Find my limit — a binary search between your highest “heard” step and the first “not heard” step above it. Each answer halves the bracket (geometrically, so it halves the musical interval) until it is narrower than about half a semitone.
• Level — the per-tone playback level, 5 % to 60 %, default 18 %. The level you used is stored with each answer.
• Burst length — 0.8 s, 1.5 s or 3 s. Longer helps in the bass, where a short burst is barely a cycle or two.
• Clear all results — wipes every answer and the fine result and returns the defaults.

Touch

• Tap a ladder cell to play it; tap the two big answer buttons to record. Every target is at least 40 px.
• Drag the level slider. It never jumps on its own.

Keyboard

• Space replay the current tone.
• Y or ↑ answer “I hear it”  ·  N or ↓ answer “I don't hear it”.
• ← → move to the previous / next ladder step and play it.
• 1 2 answer an A–B trial — the tone was in interval 1 or interval 2.
• E cycle the ear: both → left → right  ·  M switch between yes/no and A–B  ·  C toggle catch trials.
• F find my limit  ·  S start at 8 kHz  ·  R clear all results for every ear.
• Shortcuts are ignored while the level slider or any input has focus, so native slider behaviour still works. Click the widget background first.

What the ladder means

• Dashed cell with ○ — untested. Solid blue cell with ✓ — heard. Solid amber cell with ✗ — not heard. A black ring marks the current step.
• Spacing is roughly logarithmic (equal musical intervals) up to 8 kHz, then 1 kHz steps above 14 kHz, because that is where the audible ceiling usually sits.
• Every tone is a pure sine at exactly the labelled frequency, with a 30 ms fade in and a 60 ms fade out. The fades matter: a hard on/off click contains energy across the whole spectrum, so you would “hear” a 20 kHz tone that you cannot actually hear.

Tips

• Do the high end twice, once with your eyes shut. It is very easy to imagine a tone when you can see the button light up. If your two runs disagree, trust the quieter answer.
• If 20 Hz and 30 Hz are silent but 80 Hz is fine, that is almost certainly your speakers. Try headphones before you write off your low end.

Results live in this page only. Nothing is saved or sent anywhere, and reloading the page clears everything.