Average_ref_X_smooth = np.unique(np.abs(Average_ref_X_smooth))
Average_ref_X_smooth_linear = hp.linearize_pc_approx(Average_ref_X_smooth)
pos_avgpc, pos_orig_pcs, pos_norm_pcs = average_pc_waveform(
Average_ref_X_smooth, W)
pcx = interpolate.interp1d(np.linspace(0, 1, pos_avgpc.size), pos_avgpc,
"cubic")
Wp = np.zeros(n)
for i in range(Average_ref_X_smooth_linear.size - 1):
x0 = Average_ref_X_smooth_linear[i]
x1 = Average_ref_X_smooth_linear[i + 1]
Wp[x0:x1] = pcx(np.linspace(0, 1, x1 - x0))
se.save_wav(Wp, "Wp.wav", fps=fps)
'''============================================================================'''
''' PLOT LINES '''
'''============================================================================'''
fig = fig = go.Figure()
fig.layout.template = "plotly_white"
# fig.update_yaxes(zeroline=True, zerolinewidth=2, zerolinecolor="black")
fig.update_layout(
# title = name,
xaxis_title="Length",
yaxis_title="Number of Ocurrences",
# yaxis = dict( # <|<|<|<|<|<|<|<|<|<|<|<|
# scaleanchor = "x", # <|<|<|<|<|<|<|<|<|<|<|<|
# scaleratio = 1, # <|<|<|<|<|<|<|<|<|<|<|<|
# Wp[x0 : x1] = pcx(np.linspace(0, 1, x1 - x0 + 1))[0:-1]
# d_Wp[x0 : x1] = d_pcx(np.linspace(0, 1, x1 - x0 + 1))[0:-1]
Wp[x0:x1] = pcx(np.linspace(0, 1, Wp[x0:x1].size, endpoint=False))
d_Wp[x0:x1] = d_pcx(np.linspace(0, 1, d_Wp[x0:x1].size, endpoint=False))
'''==========================='''
''' Envelope: '''
'''==========================='''
Xf = np.unique(np.hstack([Xpos, Xneg]))
Ie = hp.split_raw_frontier(Xf, W, 5)
Ie = Xf[Ie].tolist()
Ae = hp.constrained_least_squares_arbitrary_intervals_X_to_Y(
Wp, d_Wp, W, Ie, 2, "k")
E = hp.coefs_to_array_arbitrary_intervals(Ae, X, Ie, n)
se.save_wav(E * Wp, f"{name}_est.wav", fps=fps)
se.save_wav(W - E * Wp, f"{name}_res.wav", fps=fps)
print("############# HERE #############")
'''====================================================================================================================='''
transp_black = "rgba(38, 12, 12, 0.2)"
def to_plot(Matrix):
X = []
Y = []
for line in Matrix:
for x, y in enumerate(line):
X.append(x)
Y.append(y)
X.append(None)
Y.append(None)
name = "piano33"
W, fps = se.read_wav(f"Samples/{name}.wav")
W = W - np.average(W)
'''get and refine frontiers '''
Xpos, Xneg = se.get_frontiers(W)
Xpos = hp.refine_frontier_iter(Xpos, W)
Xneg = hp.refine_frontier_iter(Xneg, W)
compositeW = 0
WWe = []
WWl = []
residues = [W]
for i in range(3):
We, Wl, W = loop(W, Xpos, Xneg)
compositeW += Wl
se.save_wav(Wl, f"+0{i+1}_{name}.wav", fps=fps)
WWe.append(We)
WWl.append(Wl)
residues.append(W)
se.save_wav(residues[1], "res.wav", fps=fps)
se.save_wav(compositeW, f"+Final_{name}.wav", fps=fps)
'''============================================================================'''
''' PLOT '''
'''============================================================================'''
fig = go.Figure()
fig.layout.template = "plotly_white"
# fig.update_yaxes(zeroline=True, zerolinewidth=2, zerolinecolor="black")
fig.update_layout(
# title = name,
tau = np.linalg.inv(V @ QTQinv @ V.T)
QTY = Q.T @ Y
# par1 = V @ QTQinv @ QTY
A = QTQinv @ (QTY - V.T @ tau @ V @ QTQinv @ QTY)
A = np.reshape(A, (2, -1))
print(A)
Xc = np.linspace(0, 1, avgL, dtype=np.float64)
Yc = poly.polyval(Xc, A[0, :])
ncycles = int(np.round((Xpos.size + Xneg.size) / 2))
Wr = np.tile(Yc, ncycles) * 9000
se.save_wav(Wr)
# exit()
'''============================================================================'''
''' PLOT '''
'''============================================================================'''
fig = go.Figure()
fig.layout.template = "plotly_white"
# fig.update_yaxes(zeroline=True, zerolinewidth=2, zerolinecolor="black")
fig.update_layout(
# title = name,
xaxis_title="x",
yaxis_title="Amplitude",
# yaxis = dict(scaleanchor = "x", scaleratio = 1 ), # <|<|<|<|<|<|<|<|<|<|<|<|
legend=dict(orientation='h', yanchor='top', xanchor='left', y=1.1),
margin=dict(l=5, r=5, b=5, t=5),
pa, used_positive_frontier, _ = hp.pseudocycles_average(Xpos, Xneg, W)
A = hp.approximate_pseudocycles_average(pa)
# if used_positive_frontier:
# Wp = hp.parametric_W(hp.linearize_pc(Xpos.astype(np.int)), A, n, False)
# else:
# Wp = hp.parametric_W(hp.linearize_pc(Xneg.astype(np.int)), A, n, False)
if used_positive_frontier:
Wp = hp.parametric_W(Xpos, A, n, False)
else:
Wp = hp.parametric_W(Xneg, A, n, False)
We = Wp * E
se.save_wav(We)
Xintervals = []
Yintervals = []
for x in Xf[Ix]:
Xintervals.append(x)
Xintervals.append(x)
Xintervals.append(None)
Yintervals.append(-amp)
Yintervals.append(amp)
Yintervals.append(None)
'''============================================================================'''
''' PLOT '''
'''============================================================================'''
fig = go.Figure()
fig.layout.template = "plotly_white"
# avgpc = avgpc / 7
wave = []
a = np.max(np.abs(pos_avgpc))
for i in range(Xpos.size):
amp = np.max(np.abs(pos_avgpc))
w = a * pos_avgpc / amp
a = w[-1]
for j in w:
wave.append(j)
wave = np.array(wave)
se.save_wav(wave)
'''============================================================================'''
''' PLOT '''
'''============================================================================'''
fig = go.Figure()
fig.layout.template ="plotly_white"
# fig.update_yaxes(zeroline=True, zerolinewidth=2, zerolinecolor="black")
fig.update_layout(
# title = name,
xaxis_title="x",
yaxis_title="Amplitude",
# yaxis = dict(scaleanchor = "x", scaleratio = 1 ), # <|<|<|<|<|<|<|<|<|<|<|<|
legend=dict(orientation='h', yanchor='top', xanchor='left', y=1.1),
margin=dict(l=5, r=5, b=5, t=5),
font=dict(
Xf = np.unique(np.hstack([Xpos, Xneg]))
Ix = hp.split_raw_frontier(Xf, W, 10)
A = hp.constrained_least_squares_arbitrary_intervals_wtow(
Wp, dWp, W, Xf[Ix].tolist(), 2)
E = hp.coefs_to_array_arbitrary_intervals_wtow(A, Wp, Xf[Ix].tolist(), n)
env = hp.coefs_to_array_arbitrary_intervals(A, X, Xf[Ix].tolist(), n)
# Ws = hp.parametric_W(Xpos, Ap, n, True)
We = E
# W = W * EWp
se.save_wav(We, fps=fps)
error = W - We
se.save_wav(error, "error.wav", fps=fps)
Xintervals = []
Yintervals = []
for x in Xf[Ix]:
Xintervals.append(x)
Xintervals.append(x)
Xintervals.append(None)
Yintervals.append(-amp)
Yintervals.append(amp)
Yintervals.append(None)
'''============================================================================'''
''' PLOT '''
'''============================================================================'''
Xl = np.linspace(0, 1, int(x1) - int(x0) + 1, dtype=np.float64)
Yl = poly.polyval(Xl, A)
for j in range(Yl.size - 1):
Xparam.append(x0 + j)
Wparam.append(Yl[j])
for x in range(Xp[-1], n):
Xparam.append(x)
Wparam.append(0)
Wparam = np.array(Wparam) #* amp
print(f"Wn = {W.size}, Wpn = {Wparam.size}")
# print(Wparam)
se.save_wav(Wparam)
'''============================================================================'''
''' SPLIT WAVE '''
'''============================================================================'''
n_intervals = 3
limit = np.sum(np.abs(W)) / n_intervals
Ix = []
x0 = 0
for _ in range(n_intervals - 1):
curr_sum = W[x0]
while curr_sum < limit:
x0 += 1
curr_sum += np.abs(W[x0])
Ix.append(x0 - 1)