-
Notifications
You must be signed in to change notification settings - Fork 44
/
infer_recognition_gui.py
145 lines (132 loc) · 6.55 KB
/
infer_recognition_gui.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
import argparse
import functools
import threading
import tkinter as tk
from tkinter import simpledialog
import numpy as np
import soundcard as sc
from ppvector.predict import PPVectorPredictor
from ppvector.utils.record import RecordAudio
from ppvector.utils.utils import add_arguments, print_arguments
parser = argparse.ArgumentParser(description=__doc__)
add_arg = functools.partial(add_arguments, argparser=parser)
add_arg('configs', str, 'configs/cam++.yml', '配置文件')
add_arg('use_gpu', bool, True, '是否使用GPU预测')
add_arg('audio_db_path', str, 'audio_db/', '音频库的路径')
add_arg('model_path', str, 'models/CAMPPlus_Fbank/best_model/', '导出的预测模型文件路径')
args = parser.parse_args()
print_arguments(args=args)
class VoiceRecognitionGUI:
def __init__(self, master):
master.title("夜雨飘零声纹识别")
master.geometry('430x200')
# 识别使用时间，单位秒
self.infer_time = 2
# 录音采样率
self.samplerate = 16000
# 录音块大小
self.numframes = 1024
# 模型输入长度
self.infer_len = int(self.samplerate * self.infer_time / self.numframes)
self.recognizing = False
self.record_data = []
self.record_audio = RecordAudio()
# 录音长度标签和输入框
self.record_seconds_label = tk.Label(master, text="录音长度(s):")
self.record_seconds_label.place(x=3, y=3)
self.record_seconds = tk.StringVar(value='3')
self.record_seconds_entry = tk.Entry(master, width=30, textvariable=self.record_seconds)
self.record_seconds_entry.place(x=90, y=3)
# 判断是否为同一个人的阈值标签和输入框
self.threshold_label = tk.Label(master, text="判断阈值:")
self.threshold_label.place(x=4, y=40)
self.threshold = tk.StringVar(value='0.6')
self.threshold_entry = tk.Entry(master, width=30, textvariable=self.threshold)
self.threshold_entry.place(x=90, y=40)
# 选择功能标签和按钮
self.label = tk.Label(master, text="请选择功能：")
self.label.place(x=12, y=90)
self.register_button = tk.Button(master, text="注册音频到声纹库", command=self.register)
self.register_button.place(x=90, y=90)
self.recognize_button = tk.Button(master, text="执行声纹识别", command=self.recognize)
self.recognize_button.place(x=200, y=90)
self.remove_user_button = tk.Button(master, text="删除用户", command=self.remove_user)
self.remove_user_button.place(x=290, y=90)
self.recognize_real_button = tk.Button(master, text="实时识别", command=self.recognize_thread)
self.recognize_real_button.place(x=360, y=90)
self.result_label = tk.Label(master, text="结果显示", font=('Arial', 16))
self.result_label.place(relx=0.5, y=160, anchor=tk.CENTER)
# 识别器
self.predictor = PPVectorPredictor(configs=args.configs,
threshold=float(self.threshold.get()),
audio_db_path=args.audio_db_path,
model_path=args.model_path,
use_gpu=args.use_gpu)
# 注册
def register(self):
record_seconds = int(self.record_seconds.get())
# 开始录音
self.result_label.config(text="正在录音...")
audio_data = self.record_audio.record(record_seconds=record_seconds)
self.result_label.config(text="录音结束")
name = simpledialog.askstring(title="注册", prompt="请输入注册名称")
if name is not None and name != '':
self.predictor.register(user_name=name, audio_data=audio_data, sample_rate=self.record_audio.sample_rate)
self.result_label.config(text="注册成功")
# 识别
def recognize(self):
threshold = float(self.threshold.get())
record_seconds = int(self.record_seconds.get())
# 开始录音
self.result_label.config(text="正在录音...")
audio_data = self.record_audio.record(record_seconds=record_seconds)
self.result_label.config(text="录音结束")
name, score = self.predictor.recognition(audio_data, threshold, sample_rate=self.record_audio.sample_rate)
if name:
self.result_label.config(text=f"说话人为：{name}，得分：{score}")
else:
self.result_label.config(text="没有识别到说话人，可能是没注册。")
# 删除用户
def remove_user(self):
name = simpledialog.askstring(title="删除用户", prompt="请输入删除用户名称")
if name is not None and name != '':
result = self.predictor.remove_user(user_name=name)
if result:
self.result_label.config(text="删除成功")
else:
self.result_label.config(text="删除失败")
def recognize_thread(self):
if not self.recognizing:
self.recognizing = True
self.recognize_real_button.config(text="结束声纹识别")
threading.Thread(target=self.recognize_real).start()
threading.Thread(target=self.record_real).start()
else:
self.recognizing = False
self.recognize_real_button.config(text="实时声纹识别")
# 识别
def recognize_real(self):
threshold = float(self.threshold.get())
while self.recognizing:
if len(self.record_data) < self.infer_len: continue
# 截取最新的音频数据
seg_data = self.record_data[-self.infer_len:]
audio_data = np.concatenate(seg_data)
# 删除旧的音频数据
del self.record_data[:len(self.record_data) - self.infer_len]
name, score = self.predictor.recognition(audio_data, threshold, sample_rate=self.record_audio.sample_rate)
if name:
self.result_label.config(text=f"【{name}】正在说话")
else:
self.result_label.config(text="")
def record_real(self):
self.record_data = []
default_mic = sc.default_microphone()
with default_mic.recorder(samplerate=self.samplerate, channels=1) as mic:
while self.recognizing:
data = mic.record(numframes=self.numframes)
self.record_data.append(data)
if __name__ == '__main__':
root = tk.Tk()
gui = VoiceRecognitionGUI(root)
root.mainloop()