BandHeft

BandHeft is a term used in audio signal analysis to describe the relative strength or "heft" of energy within a specified set of frequency bands. It provides a compact numeric descriptor of how much of a signal's spectral content is concentrated in the chosen bands, as opposed to being spread across the spectrum. In practice, BandHeft is computed from time-frequency representations such as the short-time Fourier transform or filter bank outputs. For a given frame, BandHeft for a band set B can be defined as the sum of the band energies within B divided by the total energy across all bands, yielding a value between 0 and 1. By adjusting B, analysts can quantify how spectral content shifts with dynamics, timbre, or spatial processing.

Applications include characterizing timbre in music information retrieval, guiding equalization, dynamic range control, audio compression, and genre or mood classification. It can be used to drive adaptive processing, such as boosting or attenuating bands with high BandHeft or stabilizing energy distribution during loud passages.

Variants and related concepts include octave-based BandHeft, Bark-scale BandHeft, or user-defined bands; related metrics include spectral centroid, spectral bandwidth, spectral entropy, and band energy ratios.

Implementation notes: Choice of band sets, window length, and normalization affects values; BandHeft is not a single standardized metric, and different studies may define it differently. See also: spectral centroid, spectral bandwidth, spectral entropy, band energy ratio.