Artificial Wasteland · research · the-pitch-that-isnt-there
The verification notebook behind The Pitch That Isn't There. The page makes a precise, checkable claim: the pitch you hear in a complex tone is the rate the whole waveform repeats — the greatest common divisor of the partial frequencies present — which is why you still hear a note after its fundamental, and every scrap of energy at it, is deleted. That is computable, not merely asserted.
Run it: node research/the-pitch-that-isnt-there/verify.mjs → 12/12 PASS. The page in public/strata/the-pitch-that-isnt-there/ synthesises every tone from the same partials the script analyses — pure sines, live Web Audio — and draws the spectrum and waveform from them as you listen.
This script proves the arithmetic — the GCD, the waveform period, and the two rival predictions. It does not measure an ear. The claim that a listener hearing the shifted complex follows the residue rather than the spacing is an experimental result (Seebeck 1841; Schouten 1940; de Boer 1956), cited — not produced here. The one experiment the page can't run for you is the one your own ear runs when you press play.
$ node research/the-pitch-that-isnt-there/verify.mjs
The Pitch That Isn't There — missing-fundamental checks
======================================================
(1) Fundamental removed — the residue still repeats at f0
partials present (Hz) : 400, 600, 800, 1000, 1200
lowest partial present : 400 Hz (no energy at or below 200 Hz)
GCD of partials : 200 Hz
waveform repetition rate : 200.21 Hz (autocorrelation, independent of the GCD math)
=> the pitch is f0 = 200 Hz, a frequency PRESENT NOWHERE in the signal
(2) Telephone band [300, 3400] Hz — the bass is reconstructed, not sent
voice fundamental : 120 Hz (and 2nd harmonic 240 Hz) — both below the band
lowest partial transmitted : 360 Hz
GCD of survivors : 120 Hz => pitch still 120 Hz
(3) de Boer's shifted complex — two models, two numbers
Δ(Hz) partials spacing→pitch residue (fine struct.) gap
0 1800/2000/2200 200.0 Hz 200.0 Hz 0.0
40 1840/2040/2240 200.0 Hz 204.0 Hz 4.0
80 1880/2080/2280 200.0 Hz 208.0 Hz 8.0
100 1900/2100/2300 200.0 Hz ambiguous (central partial at harmonic 10.5)
PASS (1) GCD of {2,3,4,5,6}·f0 equals f0
PASS (1) waveform repeats at f0 (autocorr)
PASS (2) GCD of the band-passed harmonics is f0
PASS (3) spacing (=difference tone) is invariant under Δ
PASS (3) de Boer slope = 1/n_central = 0.1 in the unambiguous band
PASS (3) ambiguity edge falls at Δ = spacing/2 (half-integer harmonic)
12/12 checks passed.