1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
|
#!/usr/bin/env python
# -*- encoding: utf-8 -*-
"""Construct resampy filters by Bayesian parameter search"""
import sys
import functools
from argparse import ArgumentParser
import numpy as np
import scipy
import scipy.signal
import optuna
import resampy
def parse_arguments(args):
parser = ArgumentParser(description=__doc__)
parser.add_argument(
"-n",
"--num-trials",
dest="n_trials",
default=1500,
help="Number of trials",
type=int,
)
parser.add_argument(
"-z",
"--num-zeros",
dest="num_zeros",
required=True,
type=int,
help="Number of zero-crossings",
)
parser.add_argument(
"--min-rolloff",
dest="min_rolloff",
default=0.9,
help="Minimum rolloff frequency (fraction of Nyquist)",
type=float,
)
parser.add_argument(
"-p",
"--precision",
dest="precision",
default=13,
help="Precision for filter interpolation",
type=int,
)
parser.add_argument(
"--min-beta",
dest="min_beta",
default=8,
help="Minimum beta for kaiser filter",
type=float,
)
parser.add_argument(
"--max-beta",
dest="max_beta",
default=20,
help="Maximum beta for kaiser filter",
type=float,
)
parser.add_argument(
"--attenuation",
dest="attenuation",
default=-120,
help="Stop-band attenuation",
type=float,
)
parser.add_argument(dest="output_file", type=str, help="Path to store filter")
return parser.parse_args(args)
def get_window(beta=None, rolloff=None, num_zeros=None):
"""Build the full window from a specification"""
win = functools.partial(scipy.signal.kaiser, beta=beta)
fil = resampy.filters.sinc_window(
num_zeros=num_zeros, precision=1, window=win, rolloff=rolloff
)[0]
# The windowed filter
fil = np.concatenate([fil[-1:0:-1], fil])
return fil
def _objective(beta, rolloff, attenuation=-120, num_zeros=40):
"""Internal objective function:
Minimize the mean passband-deviation from unity gain +
maximum stop-band gain above target attenuation
"""
fil = get_window(beta, rolloff, num_zeros=num_zeros)
W, H = scipy.signal.freqz(fil, worN=2048)
H = np.abs(H)
H_pass = H[W < 0.5 * np.pi]
H_stop = H[W >= 0.5 * np.pi]
err_pass = 20 * np.mean(np.abs(np.log10(H_pass)))
err_stop = np.max(
np.abs(np.maximum(20 * np.log10(H_stop), attenuation) - attenuation)
)
return err_pass + err_stop
def objective(trial, num_zeros, min_beta, max_beta, min_rolloff, attenuation):
"""Objective wrapper for the optimizer"""
beta = trial.suggest_float("beta", min_beta, max_beta)
rolloff = trial.suggest_float("rolloff", min_rolloff, 1.0)
loss = _objective(beta, rolloff, num_zeros=num_zeros, attenuation=attenuation)
return loss
if __name__ == "__main__":
params = parse_arguments(sys.argv[1:])
optuna.logging.set_verbosity(optuna.logging.WARNING)
sampler = optuna.samplers.TPESampler(seed=20220629)
study = optuna.create_study(direction="minimize", sampler=sampler)
func = functools.partial(
objective,
num_zeros=params.num_zeros,
min_beta=params.min_beta,
max_beta=params.max_beta,
min_rolloff=params.min_rolloff,
attenuation=params.attenuation,
)
study.optimize(func, n_trials=params.n_trials, show_progress_bar=True)
print(f"Parameters for {params.output_file}:")
print("-" * 40)
print(f"\tbeta = {study.best_params['beta']:g}")
print(f"\troll = {study.best_params['rolloff']:g}")
print(f"\t# zeros = {params.num_zeros}")
print(f"\tprecision = {params.precision}")
print(f"\tattenuation = {params.attenuation}")
print("-" * 40)
print(f"Objective value: {study.best_value:g}")
window = functools.partial(scipy.signal.kaiser, beta=study.best_params["beta"])
half_win, precision, roll = resampy.filters.sinc_window(
num_zeros=params.num_zeros,
precision=params.precision,
window=window,
rolloff=study.best_params["rolloff"],
)
np.savez(
params.output_file, half_window=half_win, precision=precision, rolloff=roll
)
|
