@audio/auditory-bark

Bark-scale critical-band filter bank (Zwicker & Terhardt 1980)

npm install @audio/auditory-bark

import barkBank from '@audio/auditory-bark'

24 Zwicker critical bands (Zwicker & Terhardt, "Analytical expressions for critical-band rate and critical bandwidth as a function of frequency", JASA 68, 1980), edges fixed at the standard table 20, 100, 200, 300, ..., 12000, 15500 Hz. Each band's center is the geometric mean of its edges, fc = sqrt(fLow*fHigh), and its coefs are a bandpass biquad at fc with Q = fc/(fHigh-fLow).

Design function, not a stream processor: barkBank(fs, opts) → bands. Positional fs first — a deliberate divergence from the filter/eq family's fn(data, params) in-place convention. No buffer in, no buffer out; it returns band descriptors. Feed each band's coefs to digital-filter's filter() to actually filter audio through it.

barkBank(44100)                                // 24 bands, defaults
barkBank(44100, { fmin: 100, fmax: 8000 })      // restrict to a sub-range

| Param | Default | | |---|---|---| | fmin | 20 | Lowest band edge Hz; a caller-supplied value below the table's first edge is honored (band 1's fLow), not clamped up | | fmax | min(15500, fs/2) | Highest band edge Hz |

Use when: fixed 24-band Bark-scale decomposition for critical-band loudness/masking models or perceptual audio coding.

Part of @audio/auditory — the auditory family umbrella.

MIT © audiojs