TestsToolsTrainLearnGear
Double-Blind Methodologyα = 0.05 Binomial Confidence

Blind ABX Tests · prove what you hear

The ABX test is the gold standard of psychoacoustics. Reference Signal A andSignal B are known. Signal X is randomly assigned to A or B. By calculating cumulative binomial probabilities across consecutive trials, our engine proves whether your detections represent genuine audible perception (p ≤ 0.05) or chance guessing.

Quick Demo: Frequency Discrimination

Signal A is 1000 Hz. Signal B is 1020 Hz (+20 Hz). Can you reliably identify X?

Perceptual Threshold Series

Choose a perceptual dimension to test your ear at different difficulty levels.

Amplitude3.0 dB down to 0.1 dB

Level Difference (ΔdB)

Test your loudness JND (Just Noticeable Difference). Can you reliably distinguish 1.0 dB, 0.5 dB, or 0.2 dB volume variations?

Take Test →
Frequency50 Hz down to 1 Hz

Frequency Difference (ΔHz)

Critical band frequency discrimination. Detect minute pitch differences from 50 Hz down to 1 Hz around a 1 kHz reference tone.

Take Test →
Musical Pitch25 cents down to 2 cents

Pitch Discrimination (Cents)

Fine musical intonation test. Discern detuned musical notes measured in cents (1/100th of a semitone) at 440 Hz.

Take Test →
Temporal40 ms down to 2 ms

Timing & Pre-delay (ms)

Temporal resolution and Haas-effect timing differences. Discriminate transient delay offsets from 40 ms down to 2 ms.

Take Test →
DynamicsDR14 vs DR10 / DR8 / DR6

Dynamic Range & Crest Factor

Subtle compression and crest factor reduction. Differentiate open dynamic recordings from soft-clipped and limited masters.

Take Test →
PhaseNormal vs 180° Inverted

Polarity Inversion (180°)

Acoustic phase inversion test. Detect whether an asymmetric waveform pushes or pulls speaker drivers on initial transient attack.

Take Test →
Digital Audio16-bit vs 8-bit to 14-bit

Bit Depth & Quantization Noise

Audibility of digital truncation. Listen for harmonic grit and quantization distortion comparing 16-bit audio against 8-bit, 10-bit, and 12-bit signals.

Take Test →

How ABX Statistical Significance Works

If you flip an unbiased coin, the probability of guessing right by sheer luck is 50% (p = 0.5). Therefore, getting 2 or 3 correct answers out of 3 proves nothing about your hearing.

Our engine computes the exact upper-tail cumulative binomial probability:

P(K ≥ k) = ∑i=kn C(n, i) × 0.5n

In rigorous scientific evaluation, we require p ≤ 0.05 (over 95% statistical confidence) to reject the null hypothesis of random guessing. This requires at minimum 5 out of 5 correct,8 out of 9, or 9 out of 10.