/
plot_wav.py
142 lines (107 loc) · 3.98 KB
/
plot_wav.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
import torch
from aecnn import AECNN
from data_io import wav_dataset, mag
from train import parse_args
import numpy as np
import os
import soundfile as sf
import matplotlib.pyplot as plt
from python_speech_features import fbank
def print_spec(array, filename, xAxisRange=None, axes='on'):
""" Print a spectrogram to a file """
if xAxisRange:
array = np.flipud(array.T)[:-3,xAxisRange[0]:xAxisRange[1]]
extent = [xAxisRange[0] / 100., xAxisRange[1] / 100., 0, 8]
else:
array = np.flipud(array.T)
extent = [0, array.shape[1] / 100., 0, 8]
fig = plt.figure()
ax = fig.add_subplot(111)
ax.imshow(array,cmap=plt.cm.jet, interpolation='none', extent=extent, aspect=1./14)
if axes == 'on':
ax.set_xlabel("Time (s)")
ax.set_ylabel("Frequency (kHz)")
fig.savefig(filename, format='pdf', bbox_inches='tight')
else:
ax.axis('off')
fig.savefig(filename, format='pdf', bbox_inches=0)
plt.close(fig)
def energy(data, window = 200):
e = np.zeros_like(data)
for i in range(len(data)-2*window):
i += window
e[i-window:i+window] += np.sum(data[i-window:i+window] ** 2) / window
cap = 0.2
e[e > cap] = cap
e[e < cap] = 0
return e
def zero_crossings(data, window = 200):
z = np.zeros_like(data)
for i in range(len(data)-2*window-1):
i += window
crossed = data[i-window:i+window] * data[i-window+1:i+window+1]
crossed[crossed > 0] = 0
crossed[crossed < 0] = 0.3
z[i-window:i+window] += np.sum(crossed) / window / window
cap = 0.2
z[z > cap] = cap
z[z < cap] = 0
return -z
def print_wav(data, fname, sr = 16000.):
e = energy(data)
z = zero_crossings(data)
fig = plt.figure()
ax = fig.add_subplot(111)
xpoints = np.arange(len(data)) / sr
ax.plot(xpoints, data, linewidth=0.5)
ax.plot(xpoints, e, linewidth=0.5)
ax.plot(xpoints, z, linewidth=0.5)
fig.savefig(fname, format='pdf', bbox_inches='tight')
plt.close(fig)
def run_test(config):
""" Define our model and test it """
generator = AECNN(
channel_counts = config.gchan,
kernel_size = config.gkernel,
block_size = config.gblocksize,
dropout = config.gdrop,
).cuda()
generator.load_state_dict(torch.load(config.gcheckpoints))
# Initialize datasets
#ev_dataset = wav_dataset(config, 'et', 4)
ev_dataset = wav_dataset(config, 'et')
#count = 0
#score = {'stoi': 0, 'estoi':0, 'sdr':0}
example = ev_dataset[361]
print(example['id'])
data = np.squeeze(generator(example['noisy'].cuda()).cpu().detach().numpy())
#clean = np.squeeze(example['clean'].numpy())
noisy = np.squeeze(example['noisy'].numpy())
#with sf.SoundFile('clean.wav', 'w', 16000, 1) as w:
# w.write(clean)
with sf.SoundFile('noisy.wav', 'w', 16000, 1) as w:
w.write(noisy)
with sf.SoundFile('test.wav', 'w', 16000, 1) as w:
w.write(data)
#print_wav(noisy, 'noisy_waveform.pdf')
#print_wav(clean, 'clean_waveform.pdf')
#print_wav(data, 'waveform.pdf')
#data = np.squeeze(generator(example['noisy']).detach().numpy())
#clean = np.squeeze(example['clean'].numpy())
#noisy = np.squeeze(example['noisy'].numpy())
#data, _ = fbank(data,nfilt=80)
#clean, _ = fbank(clean,nfilt=80)
#noisy, _ = fbank(noisy,nfilt=80)
#data, clean, noisy = np.log(data), np.log(clean), np.log(noisy)
#minimum = min(np.min(data), np.min(clean), np.min(noisy))
#data, clean, noisy = data - minimum, clean - minimum, noisy - minimum
#maximum = max(np.max(data), np.max(clean), np.max(noisy))
#data, clean, noisy = data / maximum, clean / maximum, noisy / maximum
#print_spec(data, 'spectrogram.pdf', xAxisRange=[110,140])
#print_spec(clean, 'clean_spec.pdf', xAxisRange=[110,140])
#print_spec(noisy, 'noisy_spec.pdf', xAxisRange=[110,140])
def main():
config = parse_args()
run_test(config)
if __name__=='__main__':
main()