def predlin(s,p,w):
"""
Function that computes de linear prediction using lpc for a given signal
"""
x = s*w
lpc,e = spectrum.lpc(x,N = p)
#E = np.sqrt(np.sum(x**2))
ar,sigma,lt = spectrum.aryule(x,p)
plt.subplot(3,1,1)
plt.plot(s)
plt.title("Speech signal")
xx = np.arange(0,len(w))
plt.plot(xx,w*np.max(s))
plt.subplot(312)
plt.plot(x)
plt.title("Windowed signal")
plt.subplot(313)
S_x = 20*np.log10(np.abs(np.fft.fft(x,2*len(w))))
plt.plot(S_x[0:len(w)])
ff,h = scipy.signal.freqz(1,ar,len(w),whole=False)
h2 = spectrum.arma2psd(ar,NFFT=len(w)*2)
plt.plot(20*np.log10(np.abs(h2[0:len(w)])))
def __init__(self,
xw,
dim_a=20,
dim_b=20,
max_lag=100,
NFFT=4096,
sr=44100,
Flag_Show=True):
#
self.xw = xw.copy()
self.sr = sr # sampling rate
self.NFFT = NFFT
self.Flag_Show = Flag_Show
# ARMA model, a and b
self.a, self.b, self.rho = spectrum.arma_estimate(
np.array(self.xw), dim_a, dim_b, max_lag)
if self.Flag_Show:
print('a=', self.a)
print('b=', self.b)
print('White noise variance estimate (rho) ', self.rho)
# rho_1=1.0 # if force to use 1 as rho
self.psd = spectrum.arma2psd(A=self.a,
B=self.b,
rho=self.rho,
NFFT=self.NFFT,
norm=False)
self.psd2 = self.psd[0:int(len(self.psd) / 2)] # get half side
self.get_peak()
# other ways
self.psd2_ar = None
self.psd_welch = None
def GetBurg(self, X):
import spectrum
AR, rho, ref = spectrum.arburg(X, order=20)
psd = spectrum.arma2psd(AR, rho=rho, NFFT=4096)
psd = psd[len(psd):len(psd) / 2:-1]
p = 10 * np.log(abs(psd) * 2. / (2. * np.pi))
F = spectrum.linspace(0, 80, len(p))
return p, F
def ar_model(self, ): # another way
# AR model using the Yule-Walker (autocorrelation) method
self.ar, self.var, self.reflec = spectrum.aryule(
np.array(self.xw), 30, allow_singularity=True)
if self.Flag_Show:
print('a=', self.ar)
self.psd_ar = spectrum.arma2psd(A=self.ar, NFFT=self.NFFT, norm=False)
self.psd2_ar = self.psd_ar[0:int(len(self.psd_ar) /
2)] # get half side
def __call__(self):
from spectrum import arma2psd
ar, e = modcovar(self.data, self.ar_order)
self.ar = ar
psd = arma2psd(A=ar, T=self.sampling, NFFT=self.NFFT)
if self.datatype == 'real':
if self.NFFT % 2 == 0:
newpsd = psd[0:int(self.NFFT / 2 + 1)] * 2
else:
newpsd = psd[0:int((self.NFFT + 1) / 2)] * 2
self.psd = newpsd
else:
self.psd = psd
if self.scale_by_freq is True:
self.scale()
def __call__(self):
from spectrum import arma2psd
ar, _e = arcovar(self.data, self.ar_order)
self.ar = ar
psd = arma2psd(A=ar, T=self.sampling, NFFT=self.NFFT)
if self.datatype == 'real':
if self.NFFT % 2 == 0:
newpsd = psd[0:int(self.NFFT/2+1)] * 2
else:
newpsd = psd[0:int((self.NFFT+1)/2)] * 2
self.psd = newpsd
else:
self.psd = psd
if self.scale_by_freq is True:
self.scale()
def __call__(self):
from spectrum import arma2psd
ar, e = modcovar(self.data, self.ar_order)
self.ar = ar
psd = arma2psd(A=ar, T=self.sampling, NFFT=self.NFFT)
if self.datatype == 'real':
#from .tools import twosided_2_onesided
#newpsd = twosided_2_onesided(psd)
newpsd = psd[0:int(self.NFFT//2)] * 2
# TODO: check the last value is correct or required ?
newpsd = np.append(newpsd, psd[int(self.NFFT//2)]*2)
self.psd = newpsd
else:
self.psd = psd
if self.scale_by_freq is True:
self.scale()
def __call__(self):
from spectrum import arma2psd
ar, _e = arcovar(self.data, self.ar_order)
self.ar = ar
psd = arma2psd(A=ar, T=self.sampling, NFFT=self.NFFT)
if self.datatype == 'real':
from tools import twosided_2_onesided
newpsd = twosided_2_onesided(psd)
#\psd[0:self.NFFT/2]*2
#newpsd[0] /= 2.
#newpsd = numpy.append(newpsd, psd[-1])
self.psd = newpsd
else:
self.psd = psd
if self.scale_by_freq is True:
self.scale()
def __call__(self):
from spectrum import arma2psd
ar, e = modcovar(self.data, self.ar_order)
self.ar = ar
psd = arma2psd(A=ar, T=self.sampling, NFFT=self.NFFT)
if self.datatype == 'real':
#from .tools import twosided_2_onesided
#newpsd = twosided_2_onesided(psd)
newpsd = psd[0:int(self.NFFT // 2)] * 2
# TODO: check the last value is correct or required ?
newpsd = np.append(newpsd, psd[int(self.NFFT // 2)] * 2)
self.psd = newpsd
else:
self.psd = psd
if self.scale_by_freq is True:
self.scale()
def __call__(self):
from spectrum import arma2psd
ar, e = modcovar(self.data, self.ar_order)
self.ar = ar
psd = arma2psd(A=ar, T=self.sampling, NFFT=self.NFFT)
if self.datatype == 'real':
from .tools import twosided_2_onesided
newpsd = twosided_2_onesided(psd)
#\psd[0:self.NFFT/2]*2
#newpsd[0] /= 2.
#newpsd = numpy.append(newpsd, psd[-1])
self.psd = newpsd
else:
self.psd = psd
if self.scale_by_freq is True:
self.scale()
import scipy as sp
import scipy.signal as signal
import matplotlib.pyplot as plt
import spectrum
# ### Simulo proceso AR-4
# In[2]:
M = 10000
W = np.random.randn(M)
Y = signal.lfilter([1], [1, 0.3544, 0.3508, 0.1736, 0.2401], W)
# In[3]:
psd = spectrum.arma2psd([0.3544, 0.3508, 0.1736, 0.2401], NFFT=5000)
# In[38]:
def make_plot(Pxx_den, Pxx_den_w, psd, n):
plt.figure(figsize=(16, 9))
plt.semilogy(f, (Pxx_den), label="Periodograma")
plt.semilogy(f_w, (Pxx_den_w), label="Welch")
plt.semilogy(np.linspace(0, 1, 5000), (psd), label="PSD")
plt.grid(True, which="both", alpha=0.3)
plt.ylabel("Densidad espectral [dB]", size="xx-large")
plt.xlabel("Frecuencia [rad]", size="xx-large")
plt.title(
f"Periodograma, estimador de Welch y PSD verdadera para AR-4 (n={n})",
size="xx-large")
def ar2psd( ar, n ):
'''Compute PSD from a collection of ar models'''
if ar.ndim > 1:
return np.asarray([spec.arma2psd(ar[1:,c], NFFT=2*(n-1))[:n] for c in range(ar.shape[1])]).T
else:
return spec.arma2psd(ar[1:], NFFT=2*(n-1))[:n]
def calculateFDindexes(RR, Finterp):
def power(spec,freq,fmin,fmax):
#returns power in band
band = np.array([spec[i] for i in range(len(spec)) if freq[i] >= fmin and freq[i]<fmax])
powerinband = np.sum(band)/len(spec)
return powerinband
def InterpolateRR(RR, Finterp):
# returns cubic spline interpolated array with sample rate = Finterp
step=1/Finterp
BT=np.cumsum(RR)
xmin=BT[0]
xmax=BT[-1]
BT = np.insert(BT,0,0)
BT=np.append(BT, BT[-1]+1)
RR = np.insert(RR,0,0)
RR=np.append(RR, RR[-1])
tck = interpolate.splrep(BT,RR)
BT_interp = np.arange(xmin,xmax,step)
RR_interp = interpolate.splev(BT_interp, tck)
return RR_interp, BT_interp
RR=RR/1000 #RR in seconds
RR_interp, BT_interp=InterpolateRR(RR, Finterp)
RR_interp=RR_interp-np.mean(RR_interp)
freqs=np.arange(0, 2, 0.0001)
# calculates AR coefficients
AR, P, k = spct.arburg(RR_interp*1000, 16) #burg
# estimates PSD from AR coefficients
spec = spct.arma2psd(AR, T=0.25, NFFT=2*len(freqs)) # pectrum estimation
spec = spec[0:len(spec)/2]
# WELCH psd estimation
# calculates power in different bands
VLF=power(spec,freqs,0,0.04)
LF=power(spec,freqs,0.04,0.15)
HF=power(spec,freqs,0.15,0.4)
Total=power(spec,freqs,0,2)
LFHF = LF/HF
nVLF=VLF/Total # Normalized
nLF=LF/Total
nHF=HF/Total
#NormalizedHF HFNormal
LFn=LF/(HF+LF)
HFn=HF/(HF+LF)
Power = [VLF, HF, LF]
Power_Ratio= Power/sum(Power)
# Power_Ratio=spec/sum(spec) # uncomment to calculate Spectral Entropy using all frequencies
Spectral_Entropy = 0
lenPower=0 # tengo conto delle bande che ho utilizzato
for i in xrange(0, len(Power_Ratio)):
if Power_Ratio[i]>0: # potrei avere VLF=0
Spectral_Entropy += Power_Ratio[i] * np.log(Power_Ratio[i])
lenPower +=1
Spectral_Entropy /= np.log(lenPower) #al posto di len(Power_Ratio) perche' magari non ho usato VLF
labels= np.array(['VLF', 'LF', 'HF', 'Total', 'nVLF', 'nLF', 'nHF', 'LFn', 'HFn', 'LFHF', 'SpecEn'], dtype='S10')
return [VLF, LF, HF, Total, nVLF, nLF, nHF, LFn, HFn, LFHF, Spectral_Entropy], labels
def create_all_psd():
f = pylab.linspace(0, 1, 4096)
pylab.clf()
pylab.figure(figsize=(12,8))
#MA 15 order
b, rho = spectrum.ma(data, 15, 30)
psd = spectrum.arma2psd(B=b, rho=rho)
newpsd = tools.cshift(psd, len(psd)//2) # switch positive and negative freq
pylab.plot(f, 10 * pylab.log10(newpsd/max(newpsd)), label='MA 15')
#ARMA 15 order
a, b, rho = spectrum.arma_estimate(data, 15,15, 30)
psd = spectrum.arma2psd(A=a,B=b, rho=rho)
newpsd = tools.cshift(psd, len(psd)//2) # switch positive and negative freq
pylab.plot(f, 10 * pylab.log10(newpsd/max(newpsd)), label='ARMA 15,15')
#yulewalker
ar, P,c = spectrum.aryule(data, 15, norm='biased')
psd = spectrum.arma2psd(A=ar, rho=P)
newpsd = tools.cshift(psd, len(psd)//2) # switch positive and negative freq
pylab.plot(f, 10 * pylab.log10(newpsd/max(newpsd)), label='YuleWalker 15')
#burg method
ar, P,k = spectrum.arburg(data, order=15)
psd = spectrum.arma2psd(A=ar, rho=P)
newpsd = tools.cshift(psd, len(psd)//2) # switch positive and negative freq
pylab.plot(f, 10 * pylab.log10(newpsd/max(newpsd)), label='Burg 15')
#covar method
af, pf, ab, pb, pv = spectrum.arcovar_marple(data, 15)
psd = spectrum.arma2psd(A=af, B=ab, rho=pf)
newpsd = tools.cshift(psd, len(psd)//2) # switch positive and negative freq
pylab.plot(f, 10 * pylab.log10(newpsd/max(newpsd)), label='covar 15')
#modcovar method
a, p, pv = spectrum.modcovar_marple(data, 15)
psd = spectrum.arma2psd(A=a)
newpsd = tools.cshift(psd, len(psd)//2) # switch positive and negative freq
pylab.plot(f, 10 * pylab.log10(newpsd/max(newpsd)), label='modcovar 15')
#correlogram
psd = spectrum.CORRELOGRAMPSD(data, data, lag=15)
newpsd = tools.cshift(psd, len(psd)/2) # switch positive and negative freq
pylab.plot(f, 10 * pylab.log10(newpsd/max(newpsd)), label='correlogram 15')
#minvar
psd = spectrum.minvar(data, 15)
#newpsd = tools.cshift(psd, len(psd)/2) # switch positive and negative freq
pylab.plot(f, 10 * pylab.log10(newpsd/max(newpsd)), label='MINVAR 15')
#music
psd,db = spectrum.music(data, 15, 11)
pylab.plot(f, 10 * pylab.log10(psd/max(psd)), '--',label='MUSIC 15')
#ev music
psd,db = spectrum.ev(data, 15, 11)
pylab.plot(f, 10 * pylab.log10(psd/max(psd)), '--',label='EV 15')
pylab.legend(loc='upper left', prop={'size':10}, ncol=2)
pylab.ylim([-80,10])
pylab.savefig('psd_all.png')
plt.plot(order, spectrum.AIC(len(last1second), rho, order), label='AIC')
index_min = np.argmin(spectrum.AIC(len(last1second), rho,
order)) #pick minimum rho
plt.title(index_min)
'choose just 1 second of 1 channel'
for abba in totaltrials: #for loop to do multiple second chunks
choiceofsecond = 6
last1second = channel1data[-500 * choiceofsecond:-500 *
(choiceofsecond - 1)] #pick a second chunk
last1second = last1second - mean(last1second) #remove mean
plot(last1second)
'autoregressive spectral analysis'
[ar, var, reflec] = spectrum.aryule(last1second, index_min, norm='biased')
psd = spectrum.arma2psd(ar) #power spectrum analysis
freqvals = np.linspace(0, frequency / 2, len(psd) / 2) #frequencies
#plt.plot(freqvals,psd[0:round(len(psd)/2)],label='AR')
#plt.axis([0,20,0,1500])
'fourier transform'
#ff = fft(channel1data)
#freqvals = np.fft.fftfreq(len(channel1data), 1.0/frequency)
#powerspectrumvals = abs(ff[0:round(len(ff)/2)])**2
#figure()
#plt.plot(freqvals[0:round(len(ff)/2)],powerspectrumvals,label='FFT')
#plt.axis([0,10,0,0.01])
#plt.legend()
#plt.show()
'pick frequency'
def test_covar():
af, pf, ab, pb, pv = arcovar_marple(marple_data, 15)
PSD = arma2psd(af)
newpsd = cshift(PSD, len(PSD) // 2) # switch positive and negative freq
return newpsd
"""Generate json spectrum data."""
from spectrum import arma_estimate, arma2psd, marple_data
import json
with open('dummy.json', 'w') as f:
ar, ma, rho = arma_estimate(marple_data, 15, 15, 30)
psd = arma2psd(ar, ma, rho=rho, NFFT=4096)
print(json.dumps(list(psd)), file=f)
def create_all_psd():
f = pylab.linspace(0, 1, 4096)
pylab.clf()
pylab.figure(figsize=(12, 8))
#MA 15 order
b, rho = spectrum.ma(data, 15, 30)
psd = spectrum.arma2psd(B=b, rho=rho)
newpsd = tools.cshift(psd,
len(psd) // 2) # switch positive and negative freq
pylab.plot(f, 10 * pylab.log10(newpsd / max(newpsd)), label='MA 15')
#ARMA 15 order
a, b, rho = spectrum.arma_estimate(data, 15, 15, 30)
psd = spectrum.arma2psd(A=a, B=b, rho=rho)
newpsd = tools.cshift(psd,
len(psd) // 2) # switch positive and negative freq
pylab.plot(f, 10 * pylab.log10(newpsd / max(newpsd)), label='ARMA 15,15')
#yulewalker
ar, P, c = spectrum.aryule(data, 15, norm='biased')
psd = spectrum.arma2psd(A=ar, rho=P)
newpsd = tools.cshift(psd,
len(psd) // 2) # switch positive and negative freq
pylab.plot(f,
10 * pylab.log10(newpsd / max(newpsd)),
label='YuleWalker 15')
#burg method
ar, P, k = spectrum.arburg(data, order=15)
psd = spectrum.arma2psd(A=ar, rho=P)
newpsd = tools.cshift(psd,
len(psd) // 2) # switch positive and negative freq
pylab.plot(f, 10 * pylab.log10(newpsd / max(newpsd)), label='Burg 15')
#covar method
af, pf, ab, pb, pv = spectrum.arcovar_marple(data, 15)
psd = spectrum.arma2psd(A=af, B=ab, rho=pf)
newpsd = tools.cshift(psd,
len(psd) // 2) # switch positive and negative freq
pylab.plot(f, 10 * pylab.log10(newpsd / max(newpsd)), label='covar 15')
#modcovar method
a, p, pv = spectrum.modcovar_marple(data, 15)
psd = spectrum.arma2psd(A=a)
newpsd = tools.cshift(psd,
len(psd) // 2) # switch positive and negative freq
pylab.plot(f, 10 * pylab.log10(newpsd / max(newpsd)), label='modcovar 15')
#correlogram
psd = spectrum.CORRELOGRAMPSD(data, data, lag=15)
newpsd = tools.cshift(psd,
len(psd) / 2) # switch positive and negative freq
pylab.plot(f,
10 * pylab.log10(newpsd / max(newpsd)),
label='correlogram 15')
#minvar
psd = spectrum.minvar(data, 15)
#newpsd = tools.cshift(psd, len(psd)/2) # switch positive and negative freq
pylab.plot(f, 10 * pylab.log10(newpsd / max(newpsd)), label='MINVAR 15')
#music
psd, db = spectrum.music(data, 15, 11)
pylab.plot(f, 10 * pylab.log10(psd / max(psd)), '--', label='MUSIC 15')
#ev music
psd, db = spectrum.ev(data, 15, 11)
pylab.plot(f, 10 * pylab.log10(psd / max(psd)), '--', label='EV 15')
pylab.legend(loc='upper left', prop={'size': 10}, ncol=2)
pylab.ylim([-80, 10])
pylab.savefig('psd_all.png')
def test_covar():
af, pf, ab, pb, pv = arcovar_marple(marple_data, 15)
PSD = arma2psd(af)
newpsd = cshift(PSD, len(PSD)//2) # switch positive and negative freq
return newpsd
