-
python interpreter: python3.7
-
package: SpeechRecognition pocketsphinx PyAudio PyQt5
-
IDE: PyCharm
- open settings
- open notepad
- play music
- answer her name
- say 'bye' and close
The original graphical interface was crowded, and there were no buttons for speech recognition. In order to improve the user experience, I added a button that will start speech recognition when we press it. At the same time, the identified text will be displayed.
In addition to displaying what I said, I also set up a TAB for the man-machine conversation. There is no API yet and it can only be used for simple set conversations. After that, the ChatGPT API will be plugged in for human-machine conversation.
Here's the picture of the GUI.
At the beginning, the definition of label was very confusing, so I sorted out the definition of label and redefined it to make it more readable.
At the same time, I set up a button to start and end the conversation. When the button is long pressed, the recognition begins; Recognition ends when the button is released.
Here's the code of the GUI.
class Ui_MainWindow(object):
def setupUi(self, MainWindow):
MainWindow.setObjectName("MainWindow")
self.window_width = 500
self.window_height = 700
self.label_width = self.window_width * (4 / 5)
self.label_height = self.window_height / 10
MainWindow.resize(self.window_width, self.window_height)
MainWindow.setStyleSheet("background-color: rgb(0, 0, 0);")
self.centralwidget = QtWidgets.QWidget(MainWindow)
self.centralwidget.setObjectName("centralwidget")
font = QtGui.QFont()
font.setFamily("Calibri")
font.setPointSize(12)
# 按钮控件设置
self.pushButton = QtWidgets.QPushButton(self.centralwidget)
self.pushButton.setGeometry(QtCore.QRect(self.window_width / 2 - 60, self.window_height - 100, 120, 40))
self.pushButton.setText("Start Listening")
self.pushButton.setFont(font)
self.pushButton.pressed.connect(self.start_listening)
self.pushButton.released.connect(self.stop_listening)
self.pushButton.setStyleSheet("background-color: rgb(173, 216, 230);")
# label控件设置
self.label_2 = QtWidgets.QLabel(self.centralwidget)
self.label_2.setGeometry(QtCore.QRect(60, 250, 201, 21))
self.label_3 = QtWidgets.QLabel(self.centralwidget)
self.label_4 = QtWidgets.QLabel(self.centralwidget)
self.label_5 = QtWidgets.QLabel(self.centralwidget)
self.labels = [self.label_2, self.label_3, self.label_4, self.label_5]
self.voiceFig = QtWidgets.QLabel(self.centralwidget)
self.voiceFig.setGeometry(QtCore.QRect(self.window_width / 2 - 130, 0, 100, 121))
self.voiceFig.setText("")
self.gif = QMovie("icon/voice.gif")
self.voiceFig.setMovie(self.gif)
self.gif.start()
self.voiceFig.setFixedSize(self.gif.frameRect().size() / 2.5)
self.voiceFig.setScaledContents(True)
self.voiceFig.setObjectName("voiceFig")
for idx, label in enumerate(self.labels):
label.setGeometry(QtCore.QRect(60, 280 + idx * 70, self.label_width, 50))
label.setFont(font)
label.setStyleSheet("color: rgb(0, 117, 210);")
label.setWordWrap(True)
self.label = QtWidgets.QLabel(self.centralwidget)
self.label.setGeometry(QtCore.QRect(self.window_width / 2 - 120, (self.window_height - self.label_height) / 8 + 80, self.label_width, 60))
font.setStyleStrategy(QtGui.QFont.PreferAntialias)
self.label.setFont(font)
self.label.setStyleSheet("color: rgb(0, 117, 210);")
self.label.setTextFormat(QtCore.Qt.AutoText)
self.label.setWordWrap(True)
self.label.setObjectName("label")
MainWindow.setCentralWidget(self.centralwidget)
self.statusbar = QtWidgets.QStatusBar(MainWindow)
self.statusbar.setObjectName("statusbar")
MainWindow.setStatusBar(self.statusbar)
self.retranslateUi(MainWindow)
QtCore.QMetaObject.connectSlotsByName(MainWindow)
def retranslateUi(self, MainWindow):
_translate = QtCore.QCoreApplication.translate
MainWindow.setWindowTitle(_translate("MainWindow", "Voice Assistant"))
self.label_3.setText(_translate("MainWindow", "1. Play music by saying \"Play music\""))
self.label_2.setText(_translate("MainWindow", "You can:"))
self.label.setText(_translate("MainWindow", "Hi! How can I help?"))
self.label_4.setText(_translate("MainWindow", "2. Take some notes by saying \"Open Notepad\""))
self.label_5.setText(_translate("MainWindow", " "))
def start_listening(self):
# 在这里编写开始录音的代码
print("开始录音")
def stop_listening(self):
# 在这里编写停止录音的代码
print("停止录音") Because of speech_recognition is blocking, meaning that the program will wait until the audio input is completed. You can try placing the listen method in a separate thread to avoid blocking the main thread and allow the GUI to respond normally. I use Python's built-in threading module to enable threads.
from PyQt5 import QtWidgets
from asrInterface import Ui_MainWindow
import sys
import threading
import speech_recognition as sr
import subprocess
import os
import win32api
class myWindow(QtWidgets.QMainWindow):
def __init__(self):
super(myWindow, self).__init__()
self.myCommand = " "
self.ui = Ui_MainWindow()
self.ui.setupUi(self)
self.ui.pushButton.pressed.connect(self.start_listening)
def start_listening(self):
# 开启线程
self.thread = threading.Thread(target=self.listen_audio)
self.thread.start()
def exit_program(self):
QtWidgets.QApplication.quit() # 退出程序
def listen_audio(self):
self.ui.label.setText("I'm listening:")
r = sr.Recognizer()
mic = sr.Microphone()
with mic as source:
r.adjust_for_ambient_noise(source)
audio = r.listen(source)
# 将识别结果发送到 GUI 中
self.myCommand = r.recognize_google(audio)
self.update_label()
self.myCommand = self.myCommand.lower()
if 'setting' in self.myCommand:
subprocess.Popen('start ms-settings:', shell=True)
elif 'notepad' in self.myCommand:
subprocess.Popen(['notepad.exe'], shell=True)
elif "bye" in self.myCommand:
self.exit_program()
elif 'name' in self.myCommand or 'who' in self.myCommand:
self.ui.label_5.setText("I'm your assistant, Momoko")
elif 'play music' in self.myCommand:
win32api.ShellExecute(0, 'open', 'music.wav', '', '', 1)
else:
self.ui.label_5.setText("sorry, I don't understand")
self.update_label() # 更新 label 显示
def update_label(self):
# 更新 GUI 中的 Label 显示
self.ui.label.setText(self.myCommand)
app = QtWidgets.QApplication([])
application = myWindow()
application.show()
sys.exit(app.exec()) At the beginning, I used offline speech recognition package==(r.recognize_sphinx())==, and the recognition efficiency was very low. It works because it is translated word by word, without context. After using Google api(==r.recognize_google==) to realize online real-time translation, the recognition efficiency increased by nearly 90%.
| speaking | sphinx | |
|---|---|---|
| I want to eat apple * 10 | 2/10=20% | 9/10=90% |
| open notepad for me * 10 | 1/10=10% | 8/10=80% |
| who are you * 10 | 3/10=30% | 9/10=90% |
-
As you can see from the table above, you can access an online speech recognition API to improve accuracy.
-
At the same time, recognition accuracy improves when my speech is long enough.
-
In a relatively quiet environment, in standard English.