spectral-feature

Spectral-feature concepts derive from signal processing — they are the measurable properties of signals' spectra that engineers and researchers extract and use within signal processing workflows.

spectral-feature is built and maintained for the sake of music information retrieval — MIR uses spectral features (centroid, bandwidth, rolloff, flux, slope, contrast, peaks) as its primary acoustic descriptors. The designed purpose of spectral features in audio is to serve MIR tasks.

Spectral-feature IS a specific kind of audio-feature — it extracts characteristics from the frequency-domain representation of audio (energy, slope, centroid). A competent speaker would call a spectral feature 'an audio feature.' Files against the nearest kind, establishing the rung for the ladder (Law 11e).

Spectral-envelope is a specific kind of spectral-feature: it describes the overall shape of a spectrum via smoothed amplitude values. A competent speaker would call a spectral envelope 'a spectral feature' (Law 9). Nearest kind: spectral-feature has no intermediate category between it and spectral-envelope.

Spectral-flatness IS a specific kind of spectral feature: it quantifies how noise-like vs tonal a spectrum is (Glaisher's constant / geometric-to-arithmetic mean ratio). A competent speaker would call spectral-flatness 'a spectral feature.' Nearest kind confirmed.

Spectral-bandwidth IS a specific kind of spectral-feature. It measures the width of the frequency content of a signal — a quantitative property derived from the spectral representation. A competent speaker would call bandwidth 'a spectral feature'.

A spectral peak is a specific kind of spectral feature — it is the local maximum in a frequency spectrum, representing the strongest frequency component in a given band.

Cepstral coefficients ARE a specific kind of spectral feature — they are numerical values derived from the cepstrum that represent spectral characteristics of a signal. A competent speaker would call them 'a spectral feature.' Files against nearest kind per Law 9.

Spectral-flux IS a specific kind of spectral feature: it measures the rate of change of the spectral energy over time. A competent speaker classifies it as 'a spectral feature' used in audio analysis.

Spectral-centroid IS a specific kind of spectral feature: it measures the 'center of mass' of a spectrum, representing the perceptual brightness of a sound. A competent speaker classifies it as 'a spectral feature'.

Spectral-rolloff IS a specific kind of spectral feature: it measures the frequency below which a specified percentage (e.g. 85%) of spectral energy is concentrated. A competent speaker classifies it as 'a spectral feature'.

Chroma features are a specific kind of spectral feature — they encode spectral energy distribution across 12 pitch classes. A competent speaker would call chroma features 'a spectral feature.' Files against nearest accepted kind.

Spectral slope is a specific kind of spectral feature — it quantifies the linear trend of spectral energy across frequency bins. A competent speaker would call spectral slope 'a spectral feature.' Nearest kind.

Spectral energy is a specific kind of spectral feature — it quantifies the total energy in a spectral representation. A competent speaker would call spectral energy 'a spectral feature.' Nearest kind.

Log-magnitude-spectrum is a specific kind of spectral feature — it represents spectral energy in logarithmic scale across frequency bins. A competent speaker would call a log-magnitude spectrum a spectral feature. Nearest kind.

A pitch-class-profile (PCP) is a specific kind of spectral-feature — it encodes the energy distribution across the 12 pitch classes as a 12-dimensional vector. It is a computed characteristic extracted from spectral content, not a full spectral representation. A competent speaker would call PCP a spectral feature.

Spectral peaks is a specific kind of spectral feature — a competent speaker would call it 'a spectral feature.' It identifies prominent local maxima in a magnitude spectrum, fitting squarely within the class of spectral features used in audio and signal analysis.

chroma IS a specific kind of spectral feature — it represents the distribution of acoustic energy across pitch classes (C, C#, D, ... B), computed from a frequency spectrum. A competent speaker would call chroma 'a spectral feature'. Nearest kind: spectral-feature (not stft, not magnitude-spectrum).

Spectral-contrast is a specific kind of spectral-feature: it quantifies the energy distribution across different bands of the spectrum relative to the mean. A competent speaker would call spectral contrast 'a spectral feature.'

spectral-whiteness IS a specific kind of spectral feature — it is a spectral feature that quantifies how flat the spectral density is. A competent speaker calls it 'a type of spectral feature.' Files against nearest kind per Law 9.

A mel-frequency-spectrum IS a specific kind of spectral-feature — it represents spectral content mapped to the mel-frequency scale. A competent speaker would call it 'a type of spectral feature.' Files against nearest kind (Law 9).

Spectral decay IS a specific kind of spectral feature — it characterizes the temporal evolution of a signal's spectral energy distribution. A competent speaker calls it 'a spectral feature.' The removal test is irrelevant for INSTANCE_OF; the kind-hierarchy test passes: spectral decay is a member of the class of spectral features, which are numerical descriptors extracted from a signal's spectrum.