def fbank(signal,samplerate=16000,winlen=0.025,winstep=0.01,
nfilt=26,nfft=512,lowfreq=0,highfreq=None,preemph=0.97):
"""Compute Mel-filterbank energy features from an audio signal.
:param signal: the audio signal from which to compute features. Should be an N*1 array
:param samplerate: the samplerate of the signal we are working with.
:param winlen: the length of the analysis window in seconds. Default is 0.025s (25 milliseconds)
:param winstep: the step between successive windows in seconds. Default is 0.01s (10 milliseconds)
:param nfilt: the number of filters in the filterbank, default 26.
:param nfft: the FFT size. Default is 512.
:param lowfreq: lowest band edge of mel filters. In Hz, default is 0.
:param highfreq: highest band edge of mel filters. In Hz, default is samplerate/2
:param preemph: apply preemphasis filter with preemph as coefficient. 0 is no filter. Default is 0.97.
:returns: 2 values. The first is a numpy array of size (NUMFRAMES by nfilt) containing features. Each row holds 1 feature vector. The
second return value is the energy in each frame (total energy, unwindowed)
"""
highfreq= highfreq or samplerate/2
print "preemph %s"%(preemph)
signal = sigproc.preemphasis(signal,preemph)
frames = sigproc.framesig(signal, winlen*samplerate, winstep*samplerate)
matchframes(frames[0], frames[1])
pspec = sigproc.powspec(frames,nfft)
energy = pylab.sum(pspec,1) # this stores the total energy in each frame
energy = pylab.where(energy == 0, pylab.finfo(float).eps, energy) # if energy is zero, we get problems with log
fb = get_filterbanks(nfilt, nfft, samplerate, lowfreq, highfreq)
print "len(fb) %s"%(len(fb))
colour = "k-"
for i in range(len(fb)):
if colour == "k-":
colour = "r-"
else:
colour = "k-"
startedplot = False
midpoint = 0
for j in range(len(fb[i])):
if fb[i][j] > 0:
if startedplot == False:
startedplot = j
if j > 0:
pylab.plot([j-1, j], [fb[i][j-1], fb[i][j]], colour)
if fb[i][j] == 1.0:
midpoint = j
else:
if not startedplot == False:
pylab.plot([j-1, j], [fb[i][j-1], 0], colour)
try:
print "slope to midpoint %.3f, slope from midpoint %.3f"%(1.0/float(midpoint-startedplot), 1.0/float(midpoint-j+1))
except:
pass
break
pylab.show()
feat = pylab.dot(pspec, fb.T) # compute the filterbank energies
feat = pylab.where(feat == 0, pylab.finfo(float).eps, feat) # if feat is zero, we get problems with log
return feat, energy
def ssc(signal,samplerate=16000,winlen=0.025,winstep=0.01,
nfilt=26,nfft=512,lowfreq=0,highfreq=None,preemph=0.97):
"""Compute Spectral Subband Centroid features from an audio signal.
:param signal: the audio signal from which to compute features. Should be an N*1 array
:param samplerate: the samplerate of the signal we are working with.
:param winlen: the length of the analysis window in seconds. Default is 0.025s (25 milliseconds)
:param winstep: the step between successive windows in seconds. Default is 0.01s (10 milliseconds)
:param nfilt: the number of filters in the filterbank, default 26.
:param nfft: the FFT size. Default is 512.
:param lowfreq: lowest band edge of mel filters. In Hz, default is 0.
:param highfreq: highest band edge of mel filters. In Hz, default is samplerate/2
:param preemph: apply preemphasis filter with preemph as coefficient. 0 is no filter. Default is 0.97.
:returns: A numpy array of size (NUMFRAMES by nfilt) containing features. Each row holds 1 feature vector.
"""
highfreq= highfreq or samplerate/2
signal = sigproc.preemphasis(signal,preemph)
frames = sigproc.framesig(signal, winlen*samplerate, winstep*samplerate)
pspec = sigproc.powspec(frames,nfft)
pspec = pylab.where(pspec == 0,pylab.finfo(float).eps,pspec) # if things are all zeros we get problems
fb = get_filterbanks(nfilt,nfft,samplerate,lowfreq,highfreq)
feat = pylab.dot(pspec,fb.T) # compute the filterbank energies
R = pylab.tile(pylab.linspace(1,samplerate/2,pylab.size(pspec,1)),(pylab.size(pspec,0),1))
return pylab.dot(pspec*R,fb.T) / feat
def TVD(phiOld, c, nt, limiterFunc):
"advection for nt time steps with a Courant number of c using LW"
"and upwind combined with van-Leer limiter"
# the number of independent points
nx = len(phiOld) - 1
# add two wrap-around points for cyclic boundaries
phiOld = pl.append(phiOld, [phiOld[1:3]])
# new time-step arrays for phi
phi = phiOld.copy()
# all the time steps
for it in xrange(nt):
#low and high order fluxes
phiL = phi[1:-1]
phiH = 0.5 * ((1 + c) * phi[1:-1] + (1 - c) * phi[2:])
denom = pl.where(\
(abs(phi[2:] - phi[1:-1])<pl.finfo(pl.double).resolution),\
pl.finfo(pl.double).resolution, phi[2:] - phi[1:-1])
r = (phi[1:-1] - phi[0:-2]) / denom
limiter = limiterFunc(r)
# limited flux at the mid-points
phiMid = limiter * phiH + (1 - limiter) * phiL
phi[2:-1] = phiOld[2:-1] - c * (phiMid[1:] - phiMid[:-1])
# cyclic BCs
phi[0] = phi[nx]
phi[1] = phi[nx + 1]
phi[nx + 2] = phi[2]
# update arrays
phiOld = phi.copy()
# return phiNew (without the cyclic wrap-around points)
return phi[0:nx + 1]
