Audio data accumulates fast: a single forest recorder produces tens of gigabytes per month, and a network of such recorders quickly exceeds the capacity of any human listener. Acoustic indicesare numerical summaries that map a long recording onto one or a few real numbers correlating with biological activity, spectral complexity, or anthropogenic interference. They are the workhorse of large-scale ecoacoustic monitoring.

Module 05

Acoustic Indices — Quantifying Soundscapes

Acoustic indices are summary statistics computed from audio recordings that characterize the soundscape without requiring species identification.

The Acoustic Complexity Index (ACI)

Acoustic Complexity Index \[ \text{ACI} = \sum_{k=1}^{K} \sum_{j=1}^{J-1} \frac{|I_{k,j+1} - I_{k,j}|}{\sum_{j=1}^{J} I_{k,j}} \] $I_{k,j}$ is intensity in frequency band $k$ at time step $j$. High ACI = complex, modulated soundscape typical of high biodiversity.

Acoustic Entropy Index (H)

Acoustic Entropy \[ H = H_t \cdot H_f, \qquad H_t = -\sum_{i} p_t(i)\,\log_2 p_t(i), \quad H_f = -\sum_{k} p_f(k)\,\log_2 p_f(k) \] $H \in [0,1]$: pure tones $\to 0$; broadband noise $\to 1$. Intermediate values characterise rich biological soundscapes.

The NDSI — Normalized Difference Soundscape Index

NDSI \[ \text{NDSI} = \frac{\beta - \alpha}{\beta + \alpha} \] $\alpha$ = anthrophony band (1–2 kHz), $\beta$ = biophony band (2–11 kHz). NDSI $\in [-1,+1]$: $+1$ pristine, $-1$ degraded. Analogous to NDVI in remote sensing.

Summary of Major Indices

Index	Authors	Measures	Proxy for
ACI	Pieretti 2011	Temporal intensity modulation	Biotic richness
H	Sueur 2008	Spectrotemporal entropy	Acoustic evenness
ADI	Pekin 2012	Shannon diversity	Species richness
AEI	Villanueva-Rivera 2011	Gini coefficient	Acoustic dominance
BI	Boelman 2007	Mean energy 2–8 kHz	Avian abundance
NDSI	Joo 2011	Bio/anthropo ratio	Habitat intactness

Caution

Acoustic indices are powerful proxies but not infallible. Their correlation with true species richness varies by habitat, season, and taxon. Machine learning approaches that go beyond single-number summaries are increasingly preferred.

Index correlation and complementarity

No single index captures "ecosystem health". The ACI is sensitive to short-time amplitude modulation typical of bird song; the BI weighs biological vs anthropogenic bands; the NDSI is a bio/anthrophony ratio; the H (entropy) captures spectral evenness. Using two or three complementary indices in combination dramatically outperforms any single one.

Share:X Reddit LinkedIn