def test_acorr(self):
data1 = np.sin(np.arange(1001) * 2 * np.pi / 500.)
data2 = np.sin(np.arange(1001) * 2 * np.pi / 500.)
data2[:100] = 0
data2[-100:] = 0
cor1 = xcorr.xcorr(data1, data1, 1000, padding=1)
cor2 = xcorr.acorr(data1, 1000, twosided=True, padding=1)
cor3 = xcorr.acorr(data1, 1000, twosided=True, padding=1000)
cor4 = xcorr.acorr(data1,
1000,
shift_zero=-100,
twosided=False,
padding=1)
cor5 = xcorr.acorr(data1,
1000,
shift_zero=100,
twosided=False,
padding=1)
# from pylab import plot, show, subplot, legend
# subplot(211)
# plot(data1)
# subplot(212)
# plot(cor1, label='xcorr.xcorr')
# plot(cor2, label='xcorr.acorr')
# plot(cor3, label='xcorr.acorr padding=1000')
# plot(cor4, label='xcorr.acorr shifted -100 twosided=False')
# plot(cor5, label='xcorr.acorr shifted 100 twosided=False')
# legend()
# show()
np.testing.assert_array_almost_equal(cor1, cor2)
np.testing.assert_array_almost_equal(cor1, cor3)
def test_xcorr(self):
N=1001
data1 = np.sin(np.arange(N//2+1)/100.)
data2 = np.e**(-(np.arange(N)-500)**2/100.) - np.e**(-(np.arange(N)-50)**2/100.) + 5*np.e**(-(np.arange(N)-950)**2/100.)
cor1 = scipy.signal.correlate(data1 - np.mean(data1), data2-np.mean(data2), 'full')
cor2 = xcorr.xcorr(data1, data2, 750, padding=0)
cor3 = xcorr.xcorr(data1, data2, 750, padding=1)
cor1 *= max(cor2)/max(cor1)
cor4 = scipy.signal.correlate(data2-np.mean(data2), data1 - np.mean(data1), 'full')
cor5 = xcorr.xcorr(data2, data1, 750, padding=0)
cor6 = xcorr.xcorr(data2, data1, 750, padding=1)
cor5b = xcorr.xcorr(data2, data1, 750, shift_zero=-100, padding=0)
cor5c = xcorr.xcorr(data2, data1, 750, shift_zero=100, padding=0)
cor4 *= max(cor5)/max(cor4)
cor7 = scipy.signal.correlate(data1, data2, 'full')
cor8 = xcorr.xcorr(data1, data2, 750, padding=0, demean=0)
cor7 *= max(cor8)/max(cor7)
cor9 = xcorr.xcorr(data1, data2, 750, shift_zero=100, padding=0, demean=0)
cor10 = xcorr.xcorr(data1, data2, 750, shift_zero=100, twosided=False, padding=0, demean=0)
# from pylab import plot, show, subplot, legend
# subplot(411)
# plot(data1)
# plot(data2)
# subplot(412)
# plot(cor1, label='scipy.signal demeaned')
# plot(cor2, label='xcorr.xcorr')
# plot(cor3, label='xcorr.xcorr padded to same length')
# legend()
# subplot(413)
# plot(cor4, label='scipy.signal demeaned')
# plot(cor5, label='xcorr.xcorr')
# plot(cor5b, label='xcorr.xcorr shifted -100')
# plot(cor5c, label='xcorr.xcorr shifted 100')
# plot(cor6, label='xcorr.xcorr padded to same length')
# legend()
# subplot(414)
# plot(cor7, label='scipy.signal')
# plot(cor8, label='xcorr.xcorr not demeaned')
# plot(cor9, label='xcorr.xcorr shifted 100')
# plot(cor10, label='xcorr.xcorr shifted 100 twosided=False')
# legend()
# show()
np.testing.assert_array_almost_equal(cor1, cor2)
np.testing.assert_array_almost_equal(cor4, cor5)
np.testing.assert_array_almost_equal(cor7, cor8)
np.testing.assert_array_almost_equal(cor5[200:300], cor5b[100:200])
np.testing.assert_array_almost_equal(cor5[200:300], cor5c[300:400])
def sync(c,os): aX = xcorr.xcorr( os, pilot.pilot(1) ) ret = [] for i in range( os*60 ): ret.append( aX.ce(c[i]) ) pos = ret.index(max(ret)) return ( pos, ret )
def pitch_autocorrelation(signal, sample_frequence, frame_width, shift_width,
threshold):
signal = normalization(signal)
frames = framing(signal, sample_frequence, frame_width, shift_width)
frames_energy = []
for frame in frames:
frames_energy.append(energy(frame))
fundamental_frequency_per_frame = []
for i in range(len(frames)):
if frames_energy[i] > threshold: #Voiced
try:
lags, corr = xcorr(frames[i],
maxlag=int(sample_frequence / 50))
peaks, properties = sig.find_peaks(corr)
peaks_prominences = sig.peak_prominences(signal, peaks)[0]
peaks_prominences = list(peaks_prominences)
index_max_1 = peaks_prominences.index(max(peaks_prominences))
peaks_prominences_copy = peaks_prominences
del peaks_prominences_copy[index_max_1]
index_max_2 = peaks_prominences_copy.index(
max(peaks_prominences_copy))
if index_max_2 >= index_max_1:
index_max_2 += 1
postion_max_1 = peaks[index_max_1]
postion_max_2 = peaks[index_max_2]
distance = abs(postion_max_1 - postion_max_2)
fundamental_period = distance / sample_frequence
fundamental_frequency = 1 / fundamental_period
except:
fundamental_frequency = 0
else: # Unvoiced
fundamental_frequency = 0
fundamental_frequency_per_frame.append(fundamental_frequency)
return fundamental_frequency_per_frame
def getHeartRateSchmidt(audio_data, Fs):
## Get heatrate:
# From Schmidt:
# "The duration of the heart cycle is estimated as the time from lag zero
# to the highest peaks between 500 and 2000 ms in the resulting
# autocorrelation"
# This is performed after filtering and spike removal:
Fs = float(Fs)
## 25-400Hz 4th order Butterworth band pass
audio_data = bf.butterworth_lowpass_filter(audio_data, 2, 400, Fs)
audio_data = bf.butterworth_highpass_filter(audio_data, 2, 25, Fs)
## Spike removal from the original paper:
audio_data = ssr.schmidt_spike_removal(audio_data, Fs)
## Find the homomorphic envelope
homomorphic_envelope = heh.Homomorphic_Envelope_with_Hilbert(
audio_data, Fs)
## Find the autocorrelation:
y = homomorphic_envelope - np.mean(homomorphic_envelope)
c = xcorr.xcorr(y)
signal_autocorrelation = c[len(homomorphic_envelope) + 1:]
min_index = 0.5 * Fs
max_index = 2 * Fs
index = np.argmax(signal_autocorrelation[min_index:max_index])
true_index = index + min_index - 1
heartRate = 60 / (true_index / Fs)
## Find the systolic time interval:
# From Schmidt: "The systolic duration is defined as the time from lag zero
# to the highest peak in the interval between 200 ms and half of the heart
# cycle duration"
max_sys_duration = int(np.round(((60 / heartRate) * Fs) / 2))
min_sys_duration = np.round(0.2 * Fs)
pos = np.argmax(signal_autocorrelation[min_sys_duration:max_sys_duration])
systolicTimeInterval = (min_sys_duration + pos) / Fs
return heartRate, systolicTimeInterval
def autocorrelation(frame, fs, threshold):
"""
autocorrelation(frame, fs, threshold)
Compute the pitch of frame using the autocorrelation method.
Parameters
----------
frame : ndarray
fs : float
The sampling
threshold : float
The threshold bellow which the frame is unvoiced.
Returns
-------
pitch : float
"""
if frame_energy(frame) < threshold:
return 0
lags, corr = xcorr(frame, maxlag=fs // 50)
distance = get_distance(lags, corr)
return fs / distance
def test_acorr(self):
data1 = np.sin(np.arange(1001)*2*np.pi/500.)
data2 = np.sin(np.arange(1001)*2*np.pi/500.)
data2[:100]=0
data2[-100:]=0
cor1 = xcorr.xcorr(data1, data1, 1000, padding=1)
cor2 = xcorr.acorr(data1, 1000, twosided=True, padding=1)
cor3 = xcorr.acorr(data1, 1000, twosided=True, padding=1000)
cor4 = xcorr.acorr(data1, 1000, shift_zero=-100, twosided=False, padding=1)
cor5 = xcorr.acorr(data1, 1000, shift_zero=100, twosided=False, padding=1)
# from pylab import plot, show, subplot, legend
# subplot(211)
# plot(data1)
# subplot(212)
# plot(cor1, label='xcorr.xcorr')
# plot(cor2, label='xcorr.acorr')
# plot(cor3, label='xcorr.acorr padding=1000')
# plot(cor4, label='xcorr.acorr shifted -100 twosided=False')
# plot(cor5, label='xcorr.acorr shifted 100 twosided=False')
# legend()
# show()
np.testing.assert_array_almost_equal(cor1, cor2)
np.testing.assert_array_almost_equal(cor1, cor3)
def crossCorrelate(self, a, b, maxDelay):
# maxDelay cannot be greater the length of the arrays, that will cause errors
assert maxDelay < len(a)
lags, coeffs = xcorr(a, b, maxlags=maxDelay)
return lags, coeffs
# #fr_true[pltslice, n],
# 3*fr_std[pltslice,n],
# sgax=axs[n],
# alpha=0.5)
axs[i, j].set_ylim(-0.05, 1.1)
axs[i, j].set_ylabel("$\lambda_{}(t)$".format(k + 1))
axs[i, j].set_title("Firing Rates")
axs[i, j].set_xlabel("Time")
k = k + 1
plt.tight_layout()
fig.savefig("TVpgGLM/fig/syn_tv_N10_raster.pdf")
#####################################
### Cross-correlation analysis (p3)##
#####################################
f_true = xcorr.xcorr(fr_true, dtmax=20)
f_est = xcorr.xcorr(fr_est, dtmax=20)
plt.figure()
fig, axs = plt.subplots(N, N, sharey=True)
for i in range(N):
for j in range(N):
axs[i, j].plot(f_true[i, j, :], color=color[0])
axs[i, j].plot(f_est[i, j, :], color=color[1])
axs[i, j].set_xticklabels([])
axs[i, j].set_yticklabels([])
fig.savefig("TVpgGLM/fig/syn_tv_N10_xcorr.pdf")
###################################
import xcorr import numpy as np from __microphone__ import Microphone mic1 = Microphone(0,0, 'WavFileDatabase/Data2/mic1.wav') mic2 = Microphone(0,0,'WavFileDatabase/Data2/mic2.wav') test1 = mic1.getAmplitudeList() test2 = mic2.getAmplitudeList() test1 = test1[0:50] test2 = test2[0:50] a = np.asarray(test1) b = np.asarray(test2) lags, c = xcorr.xcorr(test2, test1 ,normed=True, maxlags=49 ) print lags print c
print("5-pk")
print(GW_five_percent2)
gw_time=gw_time
gw=gw_data
ir_time2=ir_time
ir=ir_data
lpf_ir_data2=lowpass(ir,300e3)
lpf_gw_data=lowpass(gw,700e3)
plt.plot(gw_time*1e6,lpf_gw_data,ir_time[50:]*1e6,lpf_ir_data[50:])
plt.plot([gw_time[0]*1e6,gw_time[-1]*1e6],[0.05*np.max(lpf_gw_data),0.05*np.max(lpf_gw_data)],'r-')
plt.xlabel('Time ($\mu s$)')
plt.ylabel('data 2 Uncalibrated Ez from DBY station')
plt.show()
delay=xcorr(ir_time[50:],ir_time2[50:],lpf_ir_data[50:],lpf_ir_data2[50:])
#This part of the code accounts for the xcorr from 5% of GW peak
if GW_five_percent2>GW_five_percent:
xcorr=np.abs(delay)+(np.abs(GW_five_percent2-GW_five_percent))
if delay<0:
xcorr=xcorr*-1
else:
xcorr=xcorr
print("xcorr_delay")
print(xcorr)
else:
xcorr=np.abs(delay)-(np.abs(GW_five_percent2-GW_five_percent))
if delay<0:
def calculateDelay(self, x, y):
a = np.asarray(x)
b = np.asarray(y)
lags, c = xcorr(a, b, normed=True, maxlags=(len(a) - 1))
timeDelay = lags[np.argmax(c)]
return timeDelay
def test_xcorr(self):
N = 1001
data1 = np.sin(np.arange(N // 2 + 1) / 100.)
data2 = np.e**(-(np.arange(N) - 500)**2 / 100.) - np.e**(
-(np.arange(N) - 50)**2 /
100.) + 5 * np.e**(-(np.arange(N) - 950)**2 / 100.)
cor1 = scipy.signal.correlate(data1 - np.mean(data1),
data2 - np.mean(data2), 'full')
cor2 = xcorr.xcorr(data1, data2, 750, padding=0)
cor3 = xcorr.xcorr(data1, data2, 750, padding=1)
cor1 *= max(cor2) / max(cor1)
cor4 = scipy.signal.correlate(data2 - np.mean(data2),
data1 - np.mean(data1), 'full')
cor5 = xcorr.xcorr(data2, data1, 750, padding=0)
cor6 = xcorr.xcorr(data2, data1, 750, padding=1)
cor5b = xcorr.xcorr(data2, data1, 750, shift_zero=-100, padding=0)
cor5c = xcorr.xcorr(data2, data1, 750, shift_zero=100, padding=0)
cor4 *= max(cor5) / max(cor4)
cor7 = scipy.signal.correlate(data1, data2, 'full')
cor8 = xcorr.xcorr(data1, data2, 750, padding=0, demean=0)
cor7 *= max(cor8) / max(cor7)
cor9 = xcorr.xcorr(data1,
data2,
750,
shift_zero=100,
padding=0,
demean=0)
cor10 = xcorr.xcorr(data1,
data2,
750,
shift_zero=100,
twosided=False,
padding=0,
demean=0)
# from pylab import plot, show, subplot, legend
# subplot(411)
# plot(data1)
# plot(data2)
# subplot(412)
# plot(cor1, label='scipy.signal demeaned')
# plot(cor2, label='xcorr.xcorr')
# plot(cor3, label='xcorr.xcorr padded to same length')
# legend()
# subplot(413)
# plot(cor4, label='scipy.signal demeaned')
# plot(cor5, label='xcorr.xcorr')
# plot(cor5b, label='xcorr.xcorr shifted -100')
# plot(cor5c, label='xcorr.xcorr shifted 100')
# plot(cor6, label='xcorr.xcorr padded to same length')
# legend()
# subplot(414)
# plot(cor7, label='scipy.signal')
# plot(cor8, label='xcorr.xcorr not demeaned')
# plot(cor9, label='xcorr.xcorr shifted 100')
# plot(cor10, label='xcorr.xcorr shifted 100 twosided=False')
# legend()
# show()
np.testing.assert_array_almost_equal(cor1, cor2)
np.testing.assert_array_almost_equal(cor4, cor5)
np.testing.assert_array_almost_equal(cor7, cor8)
np.testing.assert_array_almost_equal(cor5[200:300], cor5b[100:200])
np.testing.assert_array_almost_equal(cor5[200:300], cor5c[300:400])
