def __init__(self, parent=None):
# open the ui
QtGui.QDialog.__init__(self, parent)
self.parent = parent
self.ui = Report.Ui_MainWindow()
self.ui.setupUi(self)
# set some parameters
self.Coords = (0, 0, 0, 0)
maxPitchLag = 3
maxVTLLag = 3
maxVarLag = 10
# build time abscisa
n = len(parent.ui.Recording)
Time = np.linspace(0, 1.0 * n / parent.ui.fs, n)
# set up waveform plot
self.ax = self.ui.RawPlot.figure.add_subplot(111)
self.ax.set_position([0.12, 0.25, 0.85, 0.63])
self.ax.plot(Time, parent.ui.Recording)
self.ax.set_title('Raw Waveform')
self.ax.set_ylabel('Amplitude')
self.ax.set_xlabel('Time (s)')
self.ax.set_xlim((0, 1.0 * n / parent.ui.fs))
self.ax.tick_params(axis='y',
which='both',
left=False,
right=False,
labelleft=False)
# set up rectangle selector widget
self.Rect = RectSelector.PersistRectangleSelector(
self.ax,
self.RawLineCallbackWrapper,
drawtype='box',
useblit=True,
button=[1, 3], # don't use middle button
minspanx=0.1,
minspany=0,
spancoords='pixels')
self.ui.RawPlot.show()
# use mean-smoothing method to calculate pitch
Pitch = np.zeros(parent.ui.PitchTime.shape)
for i in range(len(parent.ui.PitchTime)):
t = parent.ui.PitchTime[i]
RecentPitches = []
idx = 0
while i - idx >= 0 and t - parent.ui.PitchTime[i -
idx] <= maxPitchLag:
RecentPitches.append(parent.ui.Pitch[i - idx])
idx += 1
if len(RecentPitches) > 0:
Pitch[i] = np.mean(RecentPitches)
else:
Pitch[i] = np.nan
# build pitch graph
self.f0ax = self.ui.PitchPlot.figure.add_subplot(111)
self.f0ax.set_position([0.12, 0.25, 0.85, 0.63])
self.f0ax.hold(True)
self.f0ax.plot(parent.ui.Time, parent.ui.Targets[:, 0], color='aqua')
self.f0ax.plot(parent.ui.Time,
parent.ui.Targets[:, 0] + parent.ui.Targets[:, 3],
color='aqua',
linestyle='dashed')
self.f0ax.plot(parent.ui.Time,
parent.ui.Targets[:, 0] - parent.ui.Targets[:, 3],
color='aqua',
linestyle='dashed')
self.f0ax.scatter(parent.ui.PitchTime, Pitch, color='black')
self.f0ax.set_ylim((0, 500))
self.f0ax.set_xlim((0, 1.0 * n / parent.ui.fs))
self.f0ax.set_title('Fundamental Frequency')
self.f0ax.set_ylabel('F0 Frequency (Hz)')
self.f0ax.set_xlabel('Time (s)')
self.ui.PitchPlot.show()
# use mean smoothing to calculate VTL
VTL = np.zeros(parent.ui.FormantTime.shape)
for i in range(len(parent.ui.FormantTime)):
t = parent.ui.FormantTime[i]
RecentVTL = []
idx = 0
while i - idx >= 0 and t - parent.ui.FormantTime[i -
idx] <= maxVTLLag:
vtl = FormantFinder.getVocalTractLength(
parent.ui.Formants[i - idx, :], method='lammert')
if vtl > 9 and vtl < 25: # exclude implausible VTL
RecentVTL.append(vtl)
idx += 1
if len(RecentVTL) > 0:
VTL[i] = np.mean(RecentVTL)
else:
VTL[i] = np.nan
# plot VTL
self.VTLax = self.ui.VTLPlot.figure.add_subplot(111)
self.VTLax.set_position([0.12, 0.25, 0.85, 0.63])
self.VTLax.hold(True)
self.VTLax.plot(parent.ui.Time, parent.ui.Targets[:, 1], color='aqua')
self.VTLax.plot(parent.ui.Time,
parent.ui.Targets[:, 1] + parent.ui.Targets[:, 4],
color='aqua',
linestyle='dashed')
self.VTLax.plot(parent.ui.Time,
parent.ui.Targets[:, 1] - parent.ui.Targets[:, 4],
color='aqua',
linestyle='dashed')
self.VTLax.scatter(parent.ui.FormantTime, VTL, color='black')
self.VTLax.set_ylim((10, 20))
self.VTLax.set_xlim((0, 1.0 * n / parent.ui.fs))
self.VTLax.set_title('Vocal Tract Length')
self.VTLax.set_ylabel('Vocal Tract Length (cm)')
self.VTLax.set_xlabel('Time (s)')
self.ui.VTLPlot.show()
# use 10 second window to calc pitch variability
Var = np.zeros(parent.ui.PitchTime.shape)
for i in range(len(parent.ui.PitchTime)):
t = parent.ui.PitchTime[i]
RecentVar = []
idx = 0
while i - idx >= 0 and t - parent.ui.PitchTime[i -
idx] <= maxVarLag:
RecentVar.append(parent.ui.Pitch[i - idx])
idx += 1
if len(RecentVar) > 1:
m = np.mean(RecentVar)
Var[i] = np.std(
39.86 *
np.log10(np.array(RecentVar) / m)) # convert to semitones
else:
Var[i] = np.nan
# build variability plot
self.VarAx = self.ui.VarPlot.figure.add_subplot(111)
self.VarAx.set_position([0.12, 0.25, 0.85, 0.63])
self.VarAx.hold(True)
self.VarAx.plot(parent.ui.Time, parent.ui.Targets[:, 2], color='aqua')
self.VarAx.plot(parent.ui.Time,
parent.ui.Targets[:, 2] + parent.ui.Targets[:, 5],
color='aqua',
linestyle='dashed')
self.VarAx.plot(parent.ui.Time,
parent.ui.Targets[:, 2] - parent.ui.Targets[:, 5],
color='aqua',
linestyle='dashed')
self.VarAx.scatter(parent.ui.PitchTime, Var, color='black')
self.VarAx.set_ylim((0, 25))
self.VarAx.set_xlim((0, 1.0 * n / parent.ui.fs))
self.VarAx.set_title('F0 Variability')
self.VarAx.set_ylabel('F0 Variability (st)')
self.VarAx.set_xlabel('Time (s)')
self.ui.VarPlot.show()
# calculate variables for text area
D = 1.0 * n / parent.ui.fs
MP = np.nanmean(Pitch)
MVTL = np.nanmean(VTL)
MPV = np.nanmean(Var)
# set text
RecordingStats = """
Duration: %05.1f seconds
Mean Fundamental Frequency: %05.1f Hz
Mean Vocal Tract Length: %05.2f cm
Mean F0 Variability: %05.2f st
""" % (D, MP, MVTL, MPV)
self.ui.RecordingText.setText(RecordingStats)
# store variables
self.Var = Var
self.Pitch = Pitch
self.VTL = VTL
self.ui.PlayBack.clicked.connect(self._PlayCallback)
