-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtest_ok2.py
168 lines (149 loc) · 5.56 KB
/
test_ok2.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
import pandas as pd
import numpy as np
import os
import random
import sys
import glob
import json
import keras
import IPython.display as ipd
import librosa
import librosa.display
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import plotly.graph_objs as go
import plotly.offline as py
import plotly.tools as tls
import seaborn as sns
import scipy.io.wavfile
import tensorflow as tf
from keras import regularizers
from keras.callbacks import ModelCheckpoint, LearningRateScheduler, EarlyStopping
from keras.callbacks import History, ReduceLROnPlateau, CSVLogger
from keras.models import Model, Sequential
from keras.models import model_from_json
from keras.models import load_model
from keras.layers import Dense, Embedding, LSTM
from keras.layers import Input, Flatten, Dropout, Activation, BatchNormalization
from keras.layers import Conv1D, MaxPooling1D, AveragePooling1D
from keras.preprocessing import sequence
from keras.preprocessing.sequence import pad_sequences
from keras.preprocessing.text import Tokenizer
from keras.utils import np_utils
from keras.utils.np_utils import to_categorical
from sklearn.metrics import confusion_matrix
from sklearn.preprocessing import LabelEncoder
from scipy.fftpack import fft # 離散傅立葉變換 - 返回實數或複數序列
from scipy import signal
from scipy.io import wavfile
from tqdm import tqdm
from sklearn.model_selection import StratifiedShuffleSplit
from keras.models import load_model
data_test = pd.DataFrame(columns=['feature'])
input_duration=3
dir_list_test = os.listdir('voicedata/')
dir_list= os.listdir('data/')
print (dir_list)
#读取原始数据
data_df = pd.DataFrame(columns=['path', 'source', 'actor', 'gender',
'intensity', 'statement', 'repetition', 'emotion'])
data_df_test = pd.DataFrame(columns=['path','emotion'])
path="voicedata/test.mp3"
emotion=5
data_df_test.loc[0] = [path, emotion]
count = 0
for i in dir_list:
file_list = os.listdir('data/' + i)
for f in file_list:
nm = f.split('.')[0].split('-')
path = 'data/' + i + '/' + f
src = int(nm[1])
actor = int(nm[-1])
emotion = int(nm[2])
if int(actor)%2 == 0:
gender = "female"
else:
gender = "male"
if nm[3] == '01':
intensity = 0
else:
intensity = 1
if nm[4] == '01':
statement = 0
else:
statement = 1
if nm[5] == '01':
repeat = 0
else:
repeat = 1
data_df.loc[count] = [path, src, actor, gender, intensity, statement, repeat, emotion]
count += 1
label2_list = []
for i in range(len(data_df)):
if data_df.emotion[i] == 2: # Calm
lb = "_positive"
elif data_df.emotion[i] == 3: # Happy
lb = "_positive"
elif data_df.emotion[i] == 4: # Sad
lb = "_negative"
elif data_df.emotion[i] == 5: # Angry
lb = "_negative"
elif data_df.emotion[i] == 6: # Fearful
lb = "_negative"
else:
lb = "_none"
# Add gender to the label
label2_list.append(data_df.gender[i] + lb)
data_df['label'] = label2_list
data2_df = data_df.copy()
data2_df = data2_df[data2_df.label != "male_none"]
data2_df = data2_df[data2_df.label != "female_none"].reset_index(drop=True)
data2_df = data2_df[data2_df.label != "female_neutral"]
data2_df = data2_df[data2_df.label != "female_happy"]
data2_df = data2_df[data2_df.label != "female_angry"]
data2_df = data2_df[data2_df.label != "female_sad"]
data2_df = data2_df[data2_df.label != "female_fearful"]
data2_df = data2_df[data2_df.label != "female_calm"]
data2_df = data2_df[data2_df.label != "female_positive"]
data2_df = data2_df[data2_df.label != "female_negative"].reset_index(drop=True)
tmp1 = data2_df[data2_df.actor == 21]
tmp3 = data2_df[data2_df.actor == 23]
data3_df = pd.concat([tmp1, tmp3],ignore_index=True).reset_index(drop=True)
"""
for i in tqdm(range(len(data3_df))):
X, sample_rate = librosa.load(data3_df.path[i], res_type='kaiser_fast',duration=input_duration,sr=22050*2,offset=0.5)
# X = X[10000:90000]
sample_rate = np.array(sample_rate)
mfccs = np.mean(librosa.feature.mfcc(y=X, sr=sample_rate, n_mfcc=13), axis=0)
feature = mfccs
data_test.loc[i] = [feature]
"""
#使用librosa提取语音信息，保存到 data_test中
X, sample_rate = librosa.load(data_df_test.path[0], res_type='kaiser_fast',duration=input_duration,sr=22050*2,offset=0.5)
sample_rate = np.array(sample_rate)
mfccs = np.mean(librosa.feature.mfcc(y=X, sr=sample_rate, n_mfcc=13), axis=0)
feature = mfccs
data_test.loc[0] = [feature]
# evaluate 根據測試數據 test data 評估 加載的模型 loaded_model
json_file = open('model.json', 'r')
loaded_model_json = json_file.read()
json_file.close()
loaded_model = model_from_json(loaded_model_json)
# 加載 weights 到新模型中
loaded_model.load_weights("model/voiceclass.h5")
print("Loaded model from disk")
# evaluate 根據測試數據 test data 評估 加載的模型
opt = keras.optimizers.SGD(lr=0.0001, momentum=0.0, decay=0.0, nesterov=False)
loaded_model.compile(loss='categorical_crossentropy', optimizer=opt, metrics=['accuracy'])
test_valid = pd.DataFrame(data_test['feature'].values.tolist())
test_valid = np.array(test_valid)
lb = LabelEncoder()
test_valid = np.expand_dims(test_valid, axis=2)
print(test_valid)
preds = loaded_model.predict(test_valid,
batch_size=16,
verbose=1)
print(preds)
preds1=preds.argmax(axis=1)
print(preds1)