Pitch differences are easiest to detect in the mid-frequency range that covers speech and music

What it is

Pitch discrimination isn't equally fine across all frequencies. Roughly between a few hundred hertz and a few thousand hertz — the mid-frequency range that covers the human voice and most instruments' common range — the smallest frequency difference needed to tell two pitches apart is at its lowest, meaning discrimination is at its sharpest. Moving toward low frequencies (very deep tones) or high frequencies (very thin, high tones), the same discrimination task needs a larger absolute frequency gap before it's noticed, and discrimination clearly weakens. Where a pitch-based sound scheme sits in the frequency spectrum directly determines how many reliably distinguishable steps it can hold.

Why it happens

The variation in discrimination with frequency traces back again to the cochlea's place-coding mechanism: the basilar membrane's response positions for different frequencies are not evenly distributed. The mid-frequency range occupies proportionally more space along the membrane and corresponds to narrower critical bands, so the same physical distance along the membrane can resolve more frequency steps; low and high frequency ranges occupy proportionally more compressed space, so the same discrimination precision requires a larger frequency span to map onto distinguishable positions on the membrane. The mid band also happens to be where human speech's fundamental frequency and most harmonic energy concentrate, and the auditory system investing finer discrimination resources here fits with speech recognition's reliance on precise frequency information.

Studying it

The JND paradigm applies as before, but with the reference tone's frequency as the core independent variable, systematically scanned (for example, several reference points from 125 Hz to 8000 Hz), measuring the pitch JND at each point and plotting a curve of JND versus frequency. The dependent variable is the absolute JND at each frequency point, often converted to a proportion of relative frequency change for easier comparison of discrimination strength across frequency ranges. This curve is a stable, reproducible finding in auditory psychophysics, often cited directly rather than remeasured each time.

In interface design evaluation, this curve is commonly used to set a frequency-band-dependent ceiling on "how many states can be encoded with pitch" — for the same discrimination requirement, encoding in the mid-frequency range can hold more steps than encoding in the high or low range.

Where it stops holding

• This curve is measured under quiet lab conditions with single pure-tone stimuli; real environmental noise and overlapping complex timbres further compress the discrimination actually available — the lab figures are an upper-bound reference, not what's achievable in a real setting.
• Age-related high-frequency hearing loss makes discrimination in the high-frequency range degrade earlier and faster; a high-frequency discrimination curve measured on young, normal-hearing listeners doesn't transfer directly to a middle-aged or older user population.
• This curve describes pure frequency-discrimination ability, not the usability of pitch encoding in an actual product outright — low-frequency discrimination is weaker, but if only two or three coarse states need distinguishing, a low-frequency tone is still usable; weaker discrimination doesn't rule a frequency band out entirely.

Applying it

• When pitch needs to encode a larger number of states (four or more), place the encoding scheme in the mid-frequency range (roughly 500 Hz–4 kHz) to fit more reliably distinguishable steps at the same perceptual difficulty.
• If other constraints (avoiding a particular ambient-noise band, staying clear of other alarms) force the encoding into a low or high range, compensate by reducing the number of steps or widening the frequency spacing between adjacent steps, to offset that band's lower discrimination.
• How to verify it: generate a set of test tones at the designed spacing within the target frequency band, and have a sample representative of the target user population (older users included) identify each one, confirming whether the measured discrimination for that band and age group can actually support the designed number of steps.

Related

• Same group: A3.11.1 pitch is easier to discriminate than loudness · A3.11.3 absolute pitch is rare, but relative pitch comparison is nearly universal · A3.11.4 a melodic contour's rise and fall can be recognized even without an absolute pitch reference · A3.11.5 pitch encoding suits directional meaning, such as a value increasing or a state improving
• Nearby: A3.01 audible frequency range
• Search terms: frequency discrimination · pitch JND · critical band · place coding