[Python]使用SpeechRecognition進行語音辨識

本文主要使用SpeechRecognition來做一個簡單的語音辨識，使用pyqt5介面呈現。

1. 按下Start Recording，開始錄音，並顯示請開始說話。然後按鈕名改名Stop
2. 在按下Stop Recording，稍等片刻後就會呈現出辨識結果​

程式範例

import sys
import speech_recognition as sr
from PyQt5.QtWidgets import QApplication, QWidget, QPushButton, QTextEdit, QVBoxLayout
from PyQt5.QtCore import Qt
import threading

class AudioRecorder(QWidget):
    def __init__(self):
        super().__init__()

        self.initUI()
        self.recording = False
        self.recognizer = sr.Recognizer()
        self.audio = None
        self.thread = None

    def initUI(self):
        self.setGeometry(300, 300, 300, 200)
        self.setWindowTitle('Audio Recorder')

        layout = QVBoxLayout()

        self.start_button = QPushButton('Start Recording')
        self.start_button.clicked.connect(self.toggle_recording)
        layout.addWidget(self.start_button)

        self.text_area = QTextEdit()
        self.text_area.setReadOnly(True)
        layout.addWidget(self.text_area)

        self.setLayout(layout)

    def toggle_recording(self):
        if self.recording:
            self.stop_recording()
        else:
            self.start_recording()

    def start_recording(self):
        self.recording = True
        self.start_button.setText('Stop Recording')
        self.text_area.setText("請開始說話：")
        self.thread = threading.Thread(target=self.record_audio)
        self.thread.start()

    def stop_recording(self):
        self.recording = False
        self.start_button.setText('Start Recording')
        if self.thread:
            self.thread.join()
        self.recognize_audio()

    def record_audio(self):
        microphone = sr.Microphone()

        with microphone as source:
            self.recognizer.adjust_for_ambient_noise(source)
            self.audio = self.recognizer.listen(source)

    def recognize_audio(self):
        try:
            text = self.recognizer.recognize_google(self.audio, language='zh-tw')
            self.text_area.setText(text)
        except sr.UnknownValueError:
            self.text_area.setText("無法辨識音訊")
        except sr.RequestError as e:
            self.text_area.setText(f"無法連接到Google服務；{e}")

    def closeEvent(self, event):
        if self.recording:
            self.stop_recording()
        event.accept()

if __name__ == '__main__':
    app = QApplication(sys.argv)
    ex = AudioRecorder()
    ex.show()
    sys.exit(app.exec_())

程式碼重點說明

定義 AudioRecorder 類別

class AudioRecorder(QWidget):
    def __init__(self):
        super().__init__()

        self.initUI()
        self.recording = False
        self.recognizer = sr.Recognizer()
        self.audio = None
        self.thread = None

• __init__ 方法：初始化類別。這裡設置了一些初始狀態，如錄音狀態、語音辨識器和音頻數據。

錄音和停止錄音的邏輯

    def toggle_recording(self):
        if self.recording:
            self.stop_recording()
        else:
            self.start_recording()

    def start_recording(self):
        self.recording = True
        self.start_button.setText('Stop Recording')
        self.text_area.setText("請開始說話：")
        self.thread = threading.Thread(target=self.record_audio)
        self.thread.start()

    def stop_recording(self):
        self.recording = False
        self.start_button.setText('Start Recording')
        if self.thread:
            self.thread.join()
        self.recognize_audio()

• toggle_recording ：根據當前錄音狀態切換開始或停止錄音。
• start_recording ：開始錄音，將按鈕文本設為 "Stop Recording"，並啟動新執行緒來錄製音頻。
• stop_recording ：停止錄音，將按鈕文本設為 "Start Recording"，並在執行緒結束後進行語音辨識。

錄製音頻和語音辨識

    def record_audio(self):
        microphone = sr.Microphone()

        with microphone as source:
            self.recognizer.adjust_for_ambient_noise(source)
            self.audio = self.recognizer.listen(source)

    def recognize_audio(self):
        try:
            text = self.recognizer.recognize_google(self.audio, language='zh-tw')
            self.text_area.setText(text)
        except sr.UnknownValueError:
            self.text_area.setText("無法辨識音訊")
        except sr.RequestError as e:
            self.text_area.setText(f"無法連接到Google服務；{e}")

• record_audio ：使用 speech_recognition 庫來錄製音頻。這裡使用了 Microphone 來捕捉音頻，並調整麥克風的背景噪音。
• recognize_audio ：使用 Google 語音辨識服務將錄製的音頻轉換為文字，並顯示在文本區域。如果辨識失敗，則顯示錯誤信息。

視窗關閉事件

    def closeEvent(self, event):
        if self.recording:
            self.stop_recording()
        event.accept()

• closeEvent 方法：在關閉窗口時，如果仍在錄音，則停止錄音，然後接受關閉事件。