Binaural unmasking can be modelled as equalizing then subtracting two ear signals

Evidence

The equalization-cancellation model (Durlach, 1963) was designed to account for binaural signal detection in the presence of masking noise, and no attempts were made to emulate processing in the auditory pathway. In this model, the left and right inputs are first filtered with a set of band-pass filters so that the narrowband target can be more easily separated from the masker. Thereafter, the masker signal components are equalized in the two ears by adjusting the ITD and ILD values, and the ear canal signals are subtracted from each other, which ideally eliminates the masker from the signal. — Ville Pulkki, Matti Karjalainen, Communication Acoustics- An…, 1677

Categorization

• XREF: The equalization-cancellation concept is essentially a reference-channel subtraction — the same principle behind noise-cancelling headphones and adaptive beamformers; applies directly to hearing-aid algorithm design.
• DEFINE: Clear lay-accessible account of what binaural unmasking is and the EC model’s mechanism.
• SHARE: For a signal-processing colleague: Durlach’s 1963 EC model as an early precursor of adaptive noise cancellation with equalization stages.