def extract_mel_chroma_first_fft_feature(X):
X = X.astype(float)
stft = np.abs(librosa.stft(X))
chroma = np.mean(librosa.feature.chroma_stft(S=stft, sr=SAMPLE_RATE).T,
axis=0)
mel = np.mean(librosa.feature.melspectrogram(X, sr=SAMPLE_RATE).T, axis=0)
fft = np.fft.fft(X) / len(X)
fft = np.abs(fft[:len(X) // 7])
fft = [fft[0]]
welch = nk.signal_psd(X,
method="welch",
min_frequency=1,
max_frequency=20,
show=True)["Power"]
multitaper = nk.signal_psd(X,
method="multitapers",
max_frequency=20,
show=True)["Power"]
lomb = nk.signal_psd(X,
method="lomb",
min_frequency=1,
max_frequency=20,
show=True)["Power"]
welch = np.array(welch)
multitaper = np.array(multitaper)
cor = np.corrcoef(X, X)
cor = np.array([cor[0][1], cor[1][0]])
return chroma, mel, fft, welch, multitaper, cor
def extract_multitaper(X):
multitaper = nk.signal_psd(X,
method="multitapers",
max_frequency=20,
show=False)["Power"]
multitaper = np.array(multitaper)
return multitaper
def extract_welch(X):
X = X.astype(float)
welch = nk.signal_psd(X,
method="welch",
min_frequency=1,
max_frequency=20,
show=False)["Power"]
welch = np.array(welch)
return welch
def test_signal_psd(recwarn):
warnings.simplefilter("always")
data = nk.data("bio_eventrelated_100hz")
out = nk.signal_psd(data["ECG"], sampling_rate=100)
assert list(out.columns) == ["Frequency", "Power"]
assert len(recwarn) == 1
assert recwarn.pop(nk.misc.NeuroKitWarning)
def extract_burg(X):
X = X.astype(float)
burg = nk.signal_psd(X,
method="burg",
min_frequency=1,
max_frequency=20,
order=10,
show=False)["Power"]
burg = np.array(burg)
return burg
fig.savefig("README_decomposition.png", dpi=300, h_pad=3)
# =============================================================================
# Signal Power Spectrum Density
# =============================================================================
# Generate complex signal
signal = nk.signal_simulate(duration=20,
frequency=[0.5, 5, 10, 15],
amplitude=[2, 1.5, 0.5, 0.3],
noise=0.025)
# Get the PSD using different methods
welch = nk.signal_psd(signal,
method="welch",
min_frequency=1,
max_frequency=20,
show=True)
multitaper = nk.signal_psd(signal,
method="multitapers",
max_frequency=20,
show=True)
lomb = nk.signal_psd(signal,
method="lomb",
min_frequency=1,
max_frequency=20,
show=True)
burg = nk.signal_psd(signal,
method="burg",
min_frequency=1,
max_frequency=20,
def get_band(X, method):
fft = np.fft.fft(X) / len(X)
fft = np.abs(fft[:len(X) // 7])
psd_average = []
y = np.mean(
np.array(
nk.signal_psd(fft,
method=method,
min_frequency=1,
max_frequency=3,
show=False)["Power"]))
psd_average.append(y)
y = np.mean(
np.array(
nk.signal_psd(fft,
method=method,
min_frequency=4,
max_frequency=7,
show=False)["Power"]))
psd_average.append(y)
y = np.mean(
np.array(
nk.signal_psd(fft,
method=method,
min_frequency=8,
max_frequency=12,
show=False)["Power"]))
psd_average.append(y)
y = np.mean(
np.array(
nk.signal_psd(fft,
method=method,
min_frequency=13,
max_frequency=22,
show=False)["Power"]))
psd_average.append(y)
y = np.mean(
np.array(
nk.signal_psd(fft,
method=method,
min_frequency=23,
max_frequency=34,
show=False)["Power"]))
psd_average.append(y)
y = np.mean(
np.array(
nk.signal_psd(fft,
method=method,
min_frequency=35,
max_frequency=45,
show=False)["Power"]))
psd_average.append(y)
y = np.mean(
np.array(
nk.signal_psd(fft,
method=method,
min_frequency=56,
max_frequency=95,
show=False)["Power"]))
psd_average.append(y)
y = np.mean(
np.array(
nk.signal_psd(fft,
method=method,
min_frequency=105,
max_frequency=195,
show=False)["Power"]))
psd_average.append(y)
return psd_average
for database in np.unique(file["Database"]):
print(str(database))
data = file[file["Database"] == database]
for participant in np.unique(data["Participant"]):
data_participant = data[data["Participant"] == participant]
sampling_rate = np.unique(data_participant["Sampling_Rate"])[0]
rpeaks = data_participant["Rpeaks"].values
# Interpolate
rri = np.diff(rpeaks) / sampling_rate * 1000
desired_length = int(
np.rint(rpeaks[-1] / sampling_rate * sampling_rate))
rri = nk.signal_interpolate(rpeaks[1:],
rri,
x_new=np.arange(desired_length))
# Get PSD
psd = nk.signal_psd(rri, sampling_rate=sampling_rate)
#
# results = nk.hrv_frequency(rpeaks, sampling_rate=sampling_rate)
# results["Participant"] = participant
# results["Database"] = database
# results["Recording_Length"] = rpeaks[-1] / sampling_rate / 60
#
# all_results = pd.concat([all_results, results], axis=0)
#
#all_results.to_csv("data.csv", index=False)