define five acoustic indices

Author: Vincent Lostanlen

birdvoxpaint/indices.py

@@ -0,0 +1,30 @@
+import librosa
+import numpy as np
+
+
+def acoustic_complexity_index(S):
+    spectrogram = np.abs(S)**2
+    flux = np.abs(np.diff(spectrogram, axis=1))
+    return np.sum(flux, axis=1)/np.sum(spectrogram, axis=1)
+
+
+def average_energy(S):
+    spectrogram = np.abs(S)**2
+    return np.sqrt(np.mean(spectrogram, axis=1))
+
+
+def entropy_based_concentration(S):
+     spectrogram = np.abs(S)**2
+     entropy = np.array([scipy.stats.entropy(row) for row in spectrogram])
+     return (1 - entropy/np.log(spectrogram.shape[1]))
+
+
+def maximum_energy(S):
+    spectrogram = np.abs(S)**2
+    return np.max(spectrogram, axis=1)
+
+
+def maximum_flux(S):
+    spectrogram = np.abs(S)**2
+    flux = np.abs(np.diff(spectrogram, axis=1))
+    return np.max(flux, axis=1)