maad.features.acoustic_gradient_index

maad.features.acoustic_gradient_index(Sxx, dt, order=1, norm='per_bin', display=False)[source]

Compute the Acoustic Gradient Index (AGI) from a raw spectrogram.

This index must be computed on a raw spectrogram (background noise must remain).

Parameters:

Sxxndarray of floats: 2d : Spectrogram
dtfloat: Time resolution in seconds.
normstring, optional, default is ‘per_bin’: Determine if the AGI is normalized by the global mean value (‘global’ mode) or by the median value per frequency bin (‘per_bin’)

Returns:

AGI_xx2d ndarray of scalars: Acoustic Gradient Index of the spectrogram
AGI_per_bin1d ndarray of scalars: AGI value for each frequency bin sum(AGI_xx,axis=1)
AGI_sumscalar: Sum of AGI value per frequency bin (Common definition) sum(AGI_per_bin)
AGI_mean ; scalar: average AGI value per frequency bin (independant of the number of frequency bin) mean(AGI_per_bin)

Examples

>>> import maad

During the day

>>> s, fs = maad.sound.load('../data/cold_forest_daylight.wav')
>>> Sxx_power,tn,fn,_ = maad.sound.spectrogram(s,fs)
>>> _, _, AGI_mean, _ = maad.features.acoustic_gradient_index(Sxx_power,tn[1]-tn[0])
>>> AGI_mean 
5.026112548525073

During the night

>>> s, fs = maad.sound.load('../data/cold_forest_night.wav')
>>> Sxx_power,tn,fn,_ = maad.sound.spectrogram(s,fs)
>>> _, _, AGI_mean, _ = maad.features.acoustic_gradient_index(Sxx_power,tn[1]-tn[0])
>>> AGI_mean 
1.45631461307782  