def __init__(self, nfft, win_step):
def tfc(x):
return np.array([spectrogram(x[ci, :], nfft, win_step) for ci in range(x.shape[0])])
BaseNode.__init__(self)
self.nfft, self.win_step = nfft, win_step
self.n = ApplyOverInstances(tfc)
class TFC(BaseNode):
def __init__(self, nfft, win_step):
def tfc(x):
return np.array([spectrogram(x[ci, :], nfft, win_step) for ci in range(x.shape[0])])
BaseNode.__init__(self)
self.nfft, self.win_step = nfft, win_step
self.n = ApplyOverInstances(tfc)
def apply_(self, d):
assert len(d.feat_shape) == 2 # [channels x samples]
if d.feat_dim_lab is not None:
assert d.feat_dim_lab == ["channels", "time"]
tfc = self.n.apply(d)
feat_dim_lab = ["channels", "time", "frequency"]
if d.feat_lab is not None:
old_time = d.feat_lab[1]
time = np.mean(sliding_window(old_time, self.nfft, self.win_step), axis=1)
dt = np.mean(np.diff(old_time))
dt = (np.max(old_time) - np.min(old_time)) / len(old_time)
freqs = np.fft.fftfreq(self.nfft, dt)
freqs = [abs(i) for i in freqs[: self.nfft / 2 + 1]]
channels = d.feat_lab[0]
feat_lab = [channels, time.tolist(), freqs]
else:
feat_lab = None
return DataSet(feat_dim_lab=feat_dim_lab, feat_lab=feat_lab, default=tfc)