def aggregate_entropy(args):
psd = get_psd(args)
# Entropy of energy in each frequency bin over whole time
ebin = np.sum(psd, axis=1)
ebin /= np.sum(ebin)
return np.sum(-ebin * np.log2(ebin))
def spectral_decrease(args):
"""
Compute index vector
k = [0:size(X,1)-1]
k(1) = 1
kinv = 1./k
% compute slope
vsd = (kinv*(X-repmat(X(1,:),size(X,1),1)))./sum(X(2:end,:),1)
% avoid NaN for silence frames
vsd (sum(X(2:end,:),1) == 0) = 0
:param args:
:return:
"""
psd = get_psd(args)
k = np.arange(0, psd.shape[0])
k[0] = 1
kinv = 1 / k
kinv = kinv.reshape((1, psd.shape[0]))
sdecrease = np.matmul(kinv, (psd - psd[0, :])) / np.sum(psd[1:, :], axis=0)
return sdecrease
def spectral_flatness(args):
psd = get_psd(args)
nfft, noverlap = unroll_args(args, ['nfft', 'noverlap'])
hopsize = nfft - noverlap
return rosaft.spectral_flatness(y=None,
S=psd,
n_fft=nfft,
hop_length=hopsize)
def total_energy(args):
fs, nfft = unroll_args(args, ['fs', 'nfft'])
psd = get_psd(args)
# This is a little bit unclear. Eq (6.1) of Raven is the calculation below, but then it says it is in decibels,
# which this is not!
energy = np.sum(psd) * (fs / nfft)
return energy
def test_get_psd(self):
psd1, _ = _spectrogram(y=self.sig,
S=None,
n_fft=nfft,
hop_length=stepsize)
psd2 = get_psd(self.args)
self.assertTrue(np.allclose(psd1, psd2))
def spectral_crest(args):
"""
a.k.a. spectral crest factor: ratio of max freq power over mean spectrum power
:param args:
:return:
"""
psd = get_psd(args)
sc = np.max(psd, axis=0) / np.mean(psd, axis=0)
return sc
def spectral_rolloff(args):
psd = get_psd(args)
fs, nfft, noverlap = unroll_args(args, ['fs', 'nfft', 'noverlap'])
hopsize = nfft - noverlap
return rosaft.spectral_rolloff(y=None,
sr=fs,
S=psd,
n_fft=nfft,
hop_length=hopsize)
def average_entropy(args):
psd = get_psd(args)
# Entropy of each frame (time slice) averaged
newsg = (psd.T / np.sum(psd)).T
averaged_entropy = np.sum(-newsg * np.log2(newsg), axis=0)
averaged_entropy = np.mean(averaged_entropy)
return averaged_entropy
def chroma_stft(args):
psd = get_psd(args)
fs, nfft, noverlap = unroll_args(args, ['fs', 'nfft', 'noverlap'])
hopsize = nfft - noverlap
return rosaft.chroma_stft(y=None,
sr=fs,
S=psd,
n_fft=nfft,
hop_length=hopsize)
def spectral_contrast(args):
psd = get_psd(args)
fs, nfft, noverlap = unroll_args(args, ['fs', 'nfft', 'noverlap'])
hopsize = nfft - noverlap
if fs < 12800:
n_bands = 6
fmin = int(fs / 2.0**(n_bands))
else:
fmin = 200
return rosaft.spectral_contrast(y=None,
sr=fs,
S=psd,
n_fft=nfft,
hop_length=hopsize,
fmin=fmin)
def mfc(args):
psd = get_psd(args)**2
fs, nfft, ncep, fmin, fmax = unroll_args(
args, ['fs', 'nfft', ('ncep', 20), ('fmin', 0.0), ('fmax', None)])
if fmax is None:
fmax = fs // 2
# Build a Mel filter
mel_basis = _cached_get_mel_filter(sr=fs,
n_fft=nfft,
n_mels=ncep * 2,
fmin=fmin,
fmax=fmax)
melspect = np.dot(mel_basis, psd)
return power_to_db(melspect)
def spectral_flux(args):
psd = get_psd(args)
diff = np.pad(np.diff(psd), ((0, 0), (1, 0)),
'constant',
constant_values=0)
return np.linalg.norm(diff, axis=0)
def spectrum(args):
psd = get_psd(args)
return psd
def spectral_kurtosis(args):
psd = get_psd(args)
return kurtosis(psd, axis=0)
def spectral_skewness(args):
psd = get_psd(args)
return skew(psd, axis=0)
def frame_entropy(args):
psd = get_psd(args)
# Entropy of each frame (time slice) averaged
newsg = (psd.T / np.sum(psd)).T
return np.sum(-newsg * np.log2(newsg), axis=0)
