def test_againstLIBROSA_testToySig3Pure(self):
my_val = extractor.spectralSpread(signal3,
win_length=n_fft / sr,
decomposition=True)
lib_val = librosa.feature.spectral_bandwidth(y=signal3,
n_fft=n_fft,
hop_length=n_fft / 2)
corr = calculateZcorr(my_val, retrieveLibrosaValue(lib_val))
assert corr >= 0.95 # assert 95% correlation b/w zscores
def spectralSpread(self, use_librosa=False, decomposition=True,
hop_length=None, n_fft=None):
"""
Get spectral spread/bandwidth feature.
:parameters:
- use_librosa : boolean. Whether to use librosa or extractor
feature extraction function. Default False.
- decomposition : boolean. Whether to look at a time-series or
an overall root mean square analysis of the piece.
- win_length : float. Window length for the frames in the
time series analysis. [seconds]. Default is 0.05 s.
- hop_length : float. Hop length (i.e. overlap) between the
frames in the time series analysis [samples].
Default is half-overlap.
- n_fft : integer. The window length [samples]. For using with
librosa. Default 2048 [samples].
"""
if n_fft is None:
n_fft = self.n_fft
if hop_length is None:
hop_length = int(n_fft / 2)
if use_librosa:
tmp = librosa.feature.spectral_bandwidth(y=self.audio,
sr=self.sr, n_fft=n_fft, hop_length=hop_length)
self.insertFeature(tmp, 'spectralSpread', hop_length, full=False)
self.insertExtractionParameters('spectralSpread',
dict(hop_length=hop_length,
n_fft=n_fft, librosa=use_librosa,
decomposition=True))
else:
tmp = extractor.spectralSpread(self.audio,
sr=self.sr, n_fft=n_fft,
hop_length=hop_length, pad=self.pad,
decomposition=decomposition)
self.insertFeature(tmp, 'spectralSpread', hop_length, full=not decomposition)
self.insertExtractionParameters('spectralSpread',
dict(hop_length=hop_length,
n_fft=n_fft, librosa=use_librosa,
decomposition=decomposition, pad=self.pad))
return tmp
def test_againstMIR_test_alt(self):
val = extractor.spectralSpread(test_alt, sr, decomposition=False)
MIRVAL = 376.773
assert np.abs(val - MIRVAL) <= 0.15 * MIRVAL
def test_againstMIR_beethoven(self):
val = extractor.spectralSpread(beet, sr, decomposition=False)
MIRVAL = 1359.8841
assert np.abs(val - MIRVAL) <= 0.15 * MIRVAL
