forked from grausof/keras-sincnet
-
Notifications
You must be signed in to change notification settings - Fork 0
/
test.py
136 lines (113 loc) · 4.73 KB
/
test.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
from tqdm import tqdm
import soundfile as sf
import numpy as np
from conf import *
from model import *
import tensorflow as tf
import io
np.random.seed(seed)
class Validation():
def __init__(self, Batch_dev, data_folder, lab_dict, wav_lst_te, wlen, wshift, class_lay, model, debug=False):
self.wav_lst_te = wav_lst_te
self.data_folder = data_folder
self.wlen = wlen
self.wshift = wshift
self.lab_dict = lab_dict
self.Batch_dev = Batch_dev
self.class_lay = class_lay
self.model = model
self.debug = debug
def validate(self, epoch=None):
data_folder = self.data_folder
wav_lst_te = self.wav_lst_te
wlen = self.wlen
wshift = self.wshift
lab_dict = self.lab_dict
Batch_dev = self.Batch_dev
class_lay = self. class_lay
debug = self.debug
if epoch is None or epoch%N_eval_epoch==0:
print('Valuating test set...')
snt_te=len(wav_lst_te)
err_sum = 0
err_sum_snt = 0
stn_sum = 0
if debug:
print('WLEN: '+str(wlen))
print('WSHIFT: '+str(wshift))
pbar = tqdm(total=snt_te)
for i in range(snt_te):
#[signal, fs] = sf.read(data_folder+wav_lst_te[i])
fname = wav_lst_te[i]
with tf.io.gfile.GFile(fname, 'rb') as f:
[signal, fs] = sf.read(io.BytesIO(f.read()))
signal = np.array(signal)
lab_batch=lab_dict[wav_lst_te[i]]
#split signals into chunck
beg_samp=0
end_samp=wlen
N_fr=int((signal.shape[0]-wlen)/(wshift))
sig_arr=np.zeros([Batch_dev,wlen])
lab=np.zeros(N_fr+1)+lab_batch
pout = np.zeros(shape=(N_fr+1,class_lay[-1]))
count_fr=0
count_fr_tot=0
while end_samp<signal.shape[0]: #for each chunck
sig_arr[count_fr,:]=signal[beg_samp:end_samp]
beg_samp=beg_samp+wshift
end_samp=beg_samp+wlen
count_fr=count_fr+1
count_fr_tot=count_fr_tot+1
if count_fr==Batch_dev:
a,b = np.shape(sig_arr)
inp = sig_arr.reshape(a,b,1)
inp = np.array(inp)
pout[count_fr_tot-Batch_dev:count_fr_tot,:] = self.model.predict(inp, verbose=0)
count_fr=0
sig_arr=np.zeros([Batch_dev,wlen])
#Add the last items left
if count_fr>0:
inp = sig_arr[0:count_fr]
a,b = np.shape(inp)
inp = inp.reshape(a,b,1)
inp = np.array(inp)
pout[count_fr_tot-count_fr:count_fr_tot,:] = self.model.predict(inp, verbose=0)
#Prediction for each chunkc and calculation of average error
pred = np.argmax(pout, axis=1)
err = np.mean(pred!=lab)
#Calculate accuracy on the whole sentence
best_class = np.argmax(np.sum(pout, axis=0))
err_sum_snt = err_sum_snt+float((best_class!=lab[0]))
err_sum = err_sum + err
stn_sum += 1
temp_acc_stn = str(round(1-(err_sum_snt/stn_sum), 4))
temp_acc = str(round(1-(err_sum/stn_sum), 4))
if debug:
pbar.set_description('acc: {}, acc_snt: {}'.format(temp_acc, temp_acc_stn))
pbar.update(1)
#average accuracy
acc = 1-(err_sum/snt_te)
acc_snt = 1-(err_sum_snt/snt_te)
if debug:
pbar.close()
if epoch is None:
print('acc_te: {}, acc_te_snt: {}\n'.format(acc, acc_snt))
else:
print('Epoch: {}, acc_te: {}, acc_te_snt: {}\n'.format(epoch, acc, acc_snt))
with open(output_folder+"/res.res", "a") as res_file:
res_file.write("epoch %i, acc_te=%f acc_te_snt=%f\n" % (epoch, acc, acc_snt))
return (acc, acc_snt)
def main():
print("Validation...")
if pt_file!='none':
weight_file = pt_file
input_shape = (wlen, 1)
out_dim = class_lay[0]
model = getModel(input_shape, out_dim)
model.load_weights(weight_file)
val = Validation(Batch_dev, data_folder, lab_dict, wav_lst_te, wlen, wshift, class_lay, model, True)
val.validate()
else:
print("No PT FILE")
if __name__ == "__main__":
main()