Auditory Masking | WaitWhatAudio

Fully Dynamic Frequency Masking Simulator

Reposition both the frequency and amplitude of the dominant "Masker" tone. Watch the asymmetrical blue threshold wave slide across the spectrum, devouring the target Probe tone.

Masker Frequency ($f_m$): 1000 Hz

Masker Amplitude ($L_m$): 80 dB SPL

Probe Tone Frequency ($f_p$): 1200 Hz

Simultaneous Frequency Masking Threshold (Orange Peak: Variable Masker / Red Dot: Probe Tone)

Auditory Filter Width

~160 Hz

Estimated Critical Bandwidth at node

Masking Envelope Offset

42 dB SPL

Raised hearing threshold at Probe coordinate

Probe Audibility Status

MASKED (Inaudible)

Human ear psychoacoustic output state

The Layman Breakdown: Psychoacoustic Invisibility

If you whisper to someone in a quiet bedroom, they can hear you perfectly. If you whisper with the exact same volume next to a roaring subway car, your voice completely vanishes. The subway doesn't stop your vocal cords from moving—it stops your brain from processing them.

🧬

Mechanical Basilar Overload

Your inner ear processes frequencies on a tiny fluid-filled highway called the cochlea. A loud sound creates a massive physical wave that completely dominates a local section of this highway.

Smaller incoming signals landing on or near that same physical zone are completely swamped out. The mechanical neural firing triggers are simply overloaded.

🎧

The Foundation of MP3 Lossy Audio

Auditory masking is the exact reason digital audio data compression works. If a loud snare drum hit mathematically covers up a quiet hi-hat tail next to it, human ears cannot physically perceive it.

Codecs like MP3 and AAC run complex algorithms to permanently delete these "masked" audio parts entirely, shrinking file sizes by 90% without your brain ever missing a beat.

Psychoacoustic Model

Critical Bands & The Asymmetry of Masking

Human frequency selectivity is grouped into discrete, overlapping processing zones called Critical Bands. If a masker and a probe reside within the same critical band, masking is highly severe. Furthermore, masking is mathematically asymmetrical: low frequencies mask high frequencies much more efficiently than vice versa.

$$\text{Masking Threshold Shift: } \Delta L_t = f(L_{\text{masker}}, \Delta f)$$

The relationship defining the localized bandwidth profile of these internal filters can be mapped directly to the Equivalent Rectangular Bandwidth (ERB) architecture equation block:

$$\text{ERB}(f) = 24.7 \cdot (4.37 \cdot f_{\text{kHz}} + 1)$$\n $$\text{Where } \Delta f = |f_{\text{masker}} - f_{\text{probe}}| \implies \text{Masking drops exponentially as } \Delta f \to \infty$$

Upward Spread

The Frequency Slope: The masking threshold envelope tails off much slower on the high-frequency side. This means a loud 400Hz bass guitar easily swallows a quiet 600Hz guitar melody, but that same 600Hz melody barely affects the 400Hz bass line.

Temporal Bounds

Simultaneous vs. Forward Masking: While this simulation highlights simultaneous masking, masking can actually occur across time boundaries. A sudden loud explosion will hide sounds that happen up to 100-200 milliseconds *after* it stops (Forward Masking).