-
Notifications
You must be signed in to change notification settings - Fork 1
/
vis.py
205 lines (173 loc) · 7.06 KB
/
vis.py
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
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
#!/usr/bin/env python3
# Author: Armit
# Create Time: 2024/02/01
import tkinter as tk
import tkinter.ttk as ttk
import tkinter.messagebox as tkmsg
from argparse import ArgumentParser
from traceback import print_exc, format_exc
import librosa as L
import librosa.display as LD
from scipy.fftpack import fft
from noisereduce import reduce_noise
import matplotlib.pyplot as plt
from matplotlib.axes import Axes
from matplotlib.backends.backend_tkagg import FigureCanvasTkAgg
import seaborn as sns
from utils import *
COLOR_MAP = ['grey', 'r', 'g', 'b']
SPLITS = ['train', 'test1', 'test2', 'lowfid', 'highfid']
N_FFT_LIST = [2**i for i in range(3, 12)] # 8~2048
HOP_LEN_LIST = [e//2 for e in N_FFT_LIST] # 4~1024
WIN_LEN_LIST = [e//2 for e in N_FFT_LIST] # 4~1024
# defaults
SPLIT = 'train'
N_FFT = 256
HOP_LEN = 16
WIN_LEN = 64
SAMPLE_RATE_NR = 1600 # pseudo sr for noisereduce
class App:
def __init__(self, args):
self.args = args
self.X, self.Y = None, None
self.cur_idx = None
self.setup_gui()
self.setup_workspace()
try:
self.wnd.mainloop()
except KeyboardInterrupt:
self.wnd.quit()
except: print_exc()
def setup_workspace(self):
self.change_split()
def setup_gui(self):
# window
wnd = tk.Tk()
wnd.title('Signal Visualizer')
wnd.protocol('WM_DELETE_WINDOW', wnd.quit)
self.wnd = wnd
# vars
self.var_split = tk.StringVar(wnd, value=SPLIT)
self.var_idx = tk.IntVar(wnd, value=0)
self.var_n_fft = tk.IntVar(wnd, value=N_FFT)
self.var_hop_len = tk.IntVar(wnd, value=HOP_LEN)
self.var_win_len = tk.IntVar(wnd, value=WIN_LEN)
self.var_pseudo_sr = tk.IntVar(wnd, value=SAMPLE_RATE_NR)
self.var_tstart = tk.IntVar(wnd, value=0)
self.var_tend = tk.IntVar(wnd, value=NLEN-1)
# top: query
frm1 = ttk.Frame(wnd)
frm1.pack(side=tk.TOP, anchor=tk.N, expand=tk.YES, fill=tk.X)
if True:
frm11 = ttk.Frame(frm1)
frm11.pack(expand=tk.YES, fill=tk.X)
if True:
tk.Label(frm11, text='Dataset').pack(side=tk.LEFT, expand=tk.NO)
cb = ttk.Combobox(frm11, state='readonly', values=SPLITS, textvariable=self.var_split)
cb.bind('<<ComboboxSelected>>', lambda evt: self.change_split())
cb.pack(side=tk.LEFT)
tk.Label(frm11, text='FFT size').pack(side=tk.LEFT)
cb = ttk.Combobox(frm11, state='readonly', values=N_FFT_LIST, textvariable=self.var_n_fft)
cb.bind('<<ComboboxSelected>>', lambda evt: self.redraw())
cb.pack(side=tk.LEFT)
tk.Label(frm11, text='Hop length').pack(side=tk.LEFT)
cb = ttk.Combobox(frm11, state='readonly', values=HOP_LEN_LIST, textvariable=self.var_hop_len)
cb.bind('<<ComboboxSelected>>', lambda evt: self.redraw())
cb.pack(side=tk.LEFT)
tk.Label(frm11, text='Window length').pack(side=tk.LEFT)
cb = ttk.Combobox(frm11, state='readonly', values=WIN_LEN_LIST, textvariable=self.var_win_len)
cb.bind('<<ComboboxSelected>>', lambda evt: self.redraw())
cb.pack(side=tk.LEFT)
frm12 = ttk.Frame(frm1)
frm12.pack(expand=tk.YES, fill=tk.X)
if True:
tk.Label(frm12, text='Pseudo SR for NR').pack(side=tk.LEFT)
sc = tk.Scale(frm12, command=lambda _: self.redraw(), variable=self.var_pseudo_sr, orient=tk.HORIZONTAL, from_=320, to=4096, tickinterval=256, resolution=1)
sc.pack(expand=tk.YES, fill=tk.X)
self.sc = sc
frm13 = ttk.Frame(frm1)
frm13.pack(expand=tk.YES, fill=tk.X)
if True:
tk.Label(frm13, text='Sample Index').pack(side=tk.LEFT)
sc = tk.Scale(frm13, command=lambda _: self.redraw(), variable=self.var_idx, orient=tk.HORIZONTAL, from_=0, to=1000, tickinterval=500, resolution=1)
sc.pack(expand=tk.YES, fill=tk.X)
self.sc = sc
frm14 = ttk.Frame(frm1)
frm14.pack(expand=tk.YES, fill=tk.X)
if True:
tk.Label(frm14, text='Slice Left/Right').pack(side=tk.LEFT)
sc = tk.Scale(frm14, command=lambda _: self.redraw(), variable=self.var_tstart, orient=tk.HORIZONTAL, from_=0, to=NLEN-1, tickinterval=512, resolution=16)
sc.pack(side=tk.LEFT, expand=tk.YES, fill=tk.X)
sc = tk.Scale(frm14, command=lambda _: self.redraw(), variable=self.var_tend, orient=tk.HORIZONTAL, from_=0, to=NLEN-1, tickinterval=512, resolution=16)
sc.pack(side=tk.RIGHT, expand=tk.YES, fill=tk.X)
# bottom: plot
frm2 = ttk.Frame(wnd)
frm2.pack(side=tk.BOTTOM, expand=tk.YES, fill=tk.BOTH)
if True:
fig, axs = plt.subplots(5, 1, figsize=(8, 6))
fig.tight_layout()
cvs = FigureCanvasTkAgg(fig, frm2)
cvs.get_tk_widget().pack(expand=tk.YES, fill=tk.BOTH)
self.fig, self.axs, self.cvs = fig, axs, cvs
def change_split(self):
split = self.var_split.get()
if split == 'train':
self.X, self.Y = get_data_train()
elif split.endswith('fid'):
self.X = np.load(LOG_PATH / f'{split}.npy')
self.Y = get_submit_pred_maybe(len(self.X))
else:
self.X = get_data_test(split)
self.Y = get_submit_pred_maybe(len(self.X), args.fp)
nlen = len(self.X)
self.sc.config(to=nlen - 1)
self.var_idx.set(min(self.var_idx.get(), nlen - 1))
self.cur_idx = -1
self.redraw()
def redraw(self):
idx = self.var_idx .get()
n_fft = self.var_n_fft .get()
hop_len = self.var_hop_len.get()
win_len = self.var_win_len.get()
pseudo_sr = self.var_pseudo_sr.get()
tstart = self.var_tstart.get()
tend = self.var_tend.get()
if tstart >= tend:
tend = tstart + 1
if win_len >= n_fft:
self.var_win_len.set(n_fft)
win_len = n_fft
if hop_len >= n_fft:
self.var_hop_len.set(n_fft)
hop_len = n_fft
try:
x, y = self.X[idx], self.Y[idx]
x = x[tstart:tend]
try: x = reduce_noise(x, sr=pseudo_sr, n_fft=n_fft, hop_length=hop_len, win_length=win_len)
except Exception as e: print(e)
D = L.stft(x, n_fft=n_fft, hop_length=hop_len, win_length=win_len)
M = np.clip(np.log(np.abs(D) + 1e-15), a_min=1e-5, a_max=None)
c0 = L.feature.rms(y=x, frame_length=n_fft, hop_length=hop_len, pad_mode='reflect')[0]
zcr = L.feature.zero_crossing_rate(x, frame_length=n_fft, hop_length=hop_len)[0]
spec = np.abs(fft(np.expand_dims(x, axis=0), axis=1).squeeze(0))
spec = spec[1:len(spec)//8+1] # lowpass
ceps = np.abs(fft(np.expand_dims(spec, axis=0), axis=1).squeeze(0))
ceps = ceps[20:len(ceps)//4]
self.axs: List[Axes]
ax0, ax1, ax2, ax3, ax4 = self.axs
ax0.cla() ; ax0.plot(x, c=COLOR_MAP[y] if y < len(COLOR_MAP) else 'purple')
ax1.cla() ; ax1.plot(c0, label='rms') ; ax1.plot(zcr, label='zcr') ; ax1.legend(loc='upper right')
ax2.cla() ; sns.heatmap(M, ax=ax2, cbar=False) ; ax2.invert_yaxis()
ax3.cla() ; ax3.plot(spec)
ax4.cla() ; ax4.plot(ceps)
self.cvs.draw()
self.cur_idx = idx
except:
info = format_exc()
print(info)
tkmsg.showerror('Error', info)
if __name__ == '__main__':
parser = ArgumentParser()
parser.add_argument('--fp', type=Path, help='submit file')
args = parser.parse_args()
App(args)