Buffer self-check Β· mean energy per burst (measured / expected)
β
Detector must agree with the synthesis, and energy-per-burst must match independently.
If either fails, the emoji count is meaningless.
2 Β· Reference oscillator (beat test against another source)
Start this tone here, and run Szynalski or myNoise on another device nearby.
40.0 Hz vs 40.0 Hz = a steady rough tone, no pulsing.
40.0 Hz vs 40.5 Hz = a half-second warble you can count.
Use a speaker, not headphones β headphones here are silent below ~60 Hz.
Notes
Blank mode is the one to use for real sessions. A flashing phone screen in the room is an
unintended 40 Hz light stimulus β it would contaminate an audio-only experiment and
is a photosensitivity trigger.
Everything keeps running in Blank mode. Counters come from the audio clock, so they stay exact
even if the screen sleeps or the browser throttles the loop. Tap to bring them back.
Manual counting: N=10, 10 seconds β 40 groups for a 40 Hz signal (4/sec).
N=20, 10 seconds β 20 groups (2/sec) β easier to tally, same accuracy.
If a red βrow cappedβ warning appears, manual counting is invalid β read the measured rate instead.
Keyboard: Space play/stop C clear emojis Z zero counters
R reset all V cycle display mode
The MIT human protocol was a 1 ms rectangular pulse every 25 ms (4% duty cycle,
68 dB for mild-AD patients). A 40 Hz square wave is not that waveform:
50% duty, two edges per cycle (80 Hz edge rate). The sawtooth has one reset per cycle
(40 Hz) β closer to the trialed transient rate.
Neither gamma stimulation nor audio-only 40 Hz is an established treatment; HOPE
(n=663) missed its primary endpoint.