maad.features.filter_multires
- maad.features.filter_multires(Sxx, kernels, npyr=4, rescale=True)[source]
Computes 2D wavelet coefficients at multiple scales using Gaussian pyramid transformation to downscale the input spectrogram.
- Parameters:
- Sxx: list of 2D arrays
List of input spectrograms to filter
- kernels: list of 2D arrays
List of 2D kernels or filters
- npyr: int, optional
Number of pyramids to compute. Default is 4.
- rescale: boolean, optional
Indicates if the reduced images should be rescaled. Default is True.
- Returns:
- Sxx_out: list of 2D arrays
List of spectrograms filtered by each 2D kernel
Examples
>>> import maad >>> from maad.sound import load, spectrogram >>> from maad.features import filter_bank_2d_nodc, filter_multires >>> from maad import util >>> s, fs = load('../data/spinetail.wav') >>> Sxx, dt, df, ext = spectrogram(s, fs) >>> Sxx_db = util.power2dB(Sxx, db_range=80) + 80 >>> ax, fig = util.plot2d(Sxx_db, **{'extent':ext}) >>> params, kernels = filter_bank_2d_nodc(frequency=(0.5, 0.25), ntheta=2,gamma=2) >>> Sxx_out = filter_multires(Sxx, kernels, npyr=2)
Plot one of the resulting spectrograms.
>>> ax, fig = util.plot2d(Sxx_out[5], **{'extent':ext})