maad.features.tfsd

maad.features.tfsd(Sxx, fn, tn, flim=(2000, 8000), log=True, mode='thirdOctave', display=False)[source]

Compute the Time frequency derivation index (tfsd) from a spectrogram. [1] [2]

Parameters:

Sxxndarray of floats: matrix : Amplitude spectrogram
fnvector: frequency vector corresponding to the spectrogram
tnvector: time vector corresponding to the spectrogram
flimtupple (fmin, fmax), optional, default is (2000, 8000): Frequency band used to compute tfsd.
logboolean, optional, default is True: If True, the log of Sxx is taken before calculating the TFSD as it was defined in [1] [2] if False, no log is applied, TFSD > 0.3 indicates the presence of birds
modestring {‘thirdOctave’,’Octave’}, default is thirdOctave: Select the way to transform the spectrogram with linear bands into octave bands
displayboolean, optional, default is False: Display the 1st and 2nd derivation of the spectrogram

Returns:

tfsdscalar: Time frequency derivation index

Notes

The TFSD varies between 0 and 1, where 1 indicates sound events on the full spectrogram. When flim=(2000,8000), TFSD mostly indicates the presence of birds in the signal When flim=(0,1000), TFSD mostly indicates the presence of human voice in the signal When log=False and flim=(2000,8000), a TFSD > 0.3 indicates the presence of birds in the signal

References

[1] (1,2)

Aumond, P., Can, A., De Coensel, B., Botteldooren, D., Ribeiro, C., & Lavandier, C. (2017). Modeling soundscape pleasantness using perceptual assessments and acoustic measurements along paths in urban context. Acta Acustica united with Acustica. DOI: 10.3813/AAA.919073

[2] (1,2)

Gontier, F., Lavandier, C., Aumond, P., Lagrange, M., & Petiot, J. F. (2019). Estimation of the perceived time of presence of sources in urban acoustic environments using deep learning techniques. Acta Acustica united with Acustica.`DOI: 10.3813/AAA.919384 <https://doi.org/10.3813/AAA.919384>`_

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)
>>> maad.features.tfsd(Sxx_power,fn, tn)  
0.4637352908292768

During the night

>>> s, fs = maad.sound.load('../data/cold_forest_night.wav')
>>> Sxx_power,tn,fn,_ = maad.sound.spectrogram(s,fs)
>>> maad.features.tfsd(Sxx_power,fn, tn)   
0.42435773298811824