ITAコーパスの文章リストを画面に表示させ連続で録音するアプリ。

pip install SpeechRecognition

import tkinter as tk
from tkinter import scrolledtext as tk_scrolledtext
import time
from tkinter import messagebox
import os
from pydub import AudioSegment
from pydub.playback import play

import threading

import speech_recognition as sr
import re
SAMPLE_RATE = 44100 

# 録音スレッドの処理
def record_audio():
    global recording 
    global lines    
    global ix
    length = len(lines)
    r = sr.Recognizer()
    while recording:  # recordingがTrueの間、録音を継続
        if len(listbox.curselection()) == 0:
            ix = None
        else:
            ix = listbox.curselection()[0]
        if ix  is None:
            record_button.config(text="Start Recording")
            stop_recording()
            output_text.insert(tk.END, 'stop_recording\n')
            break

        print('record_audio')
        line2 = re.split(':|,|\n', lines[ix])
        print("\007")
        time.sleep(1.5)
        output_text.delete('1.0','end'	)
        print( f'{ix+1}/{length}\n')
        print( f'{line2[0]}\n')
        output_text.insert(tk.END, f'{line2[0]}\n')
        if len(line2) > 1:
            print( f'{line2[1]}\n')
            output_text.insert(tk.END, f'{line2[1]}\n')

        with sr.Microphone(sample_rate=SAMPLE_RATE) as source:
            r.dynamic_energy_threshold = True  # type: bool
            r.adjust_for_ambient_noise(source, duration=0.2)
            print("[なにか話してください]")
            audio = r.listen(source)
#

        print("音声処理中 ...")
        output_text.insert(tk.END, '音声処理中 ...\n')
# write audio to a WAV file
        output_filename = dir_path+"/"+line2[0]+".wav"  # 出力ファイル名を固定化
        sound = AudioSegment(audio.get_raw_data(), sample_width=2, channels=1, frame_rate=SAMPLE_RATE)
        newsound = sound + 3
        soundtime = newsound.duration_seconds # 再生時間(秒)

        print( f'{ix+1}/{length} 録音時間(秒)：{soundtime}\n')
        output_text.insert(tk.END, f'{ix+1}/{length} 録音時間(秒)：{soundtime}\n')
        shift_jis_text = output_filename.encode('shift_jis').decode('shift_jis')

        print(shift_jis_text)  # 変換された文字列の表示

        file_handle = newsound.export(shift_jis_text, format="wav")

#        with open(output_filename, "wb") as f:
#            f.write(audio.get_wav_data())
#
        if check_val.get():
        # 生成された音声を再生
            print("音声再生")
            output_text.insert(tk.END, '音声再生\n')
            play(newsound)
            print("音声再生完了")
            output_text.insert(tk.END, '音声再生完了\n')
        print("音声処理完了")
        output_text.insert(tk.END, '音声処理完了\n')
#
        listbox.select_clear(0, tk.END)
        time.sleep(3)

        ix +=1
        if length <= ix:
            record_button.config(text="Start Recording")
            stop_recording()
            output_text.insert(tk.END, 'stop_recording\n')
            break

        listbox.selection_set(ix) 
        listbox.see(ix)

"""
"""

# 
def start_recording():
    global recording
    global starting
    global th
    
    recording = True
    starting = True
    th = threading.Thread(target=record_audio)
    th.start()
    print("start_recording ...")

# 録音を停止する関数
def stop_recording():
    global recording
    recording = False
#    th.join()
    print("stop_recording ...")

# 録音ボタンがクリックされたときの処理
def toggle_recording():
    global recording
    if recording:
        # 録音を停止
        stop_recording()
        record_button.config(text="Start Recording")
    else:
        # 録音を開始
        start_recording()
        record_button.config(text="Stop Recording")

def delete_window():
    global recording
    global th
    global starting
       
    # 録音を停止
    record_button.config(text="Start Recording")
    record_button["state"] = "disabled"
    stop_recording()
  
    if   not starting :
        root.destroy()
    elif   not th.is_alive() :
        root.destroy()
    else:
        messagebox.showerror('error', 'running. Wait a minute')

def read_file(file_path):
    with open(file_path, 'r', encoding='utf-8', errors='ignore')  as file:  # ファイルをUTF-8エンコーディングで読み込み
        return file.readlines()


if __name__ == "__main__":
#
    file_path = "recitation_transcript_utf8.txt"  # 入力ファイル名を固定化
#    file_path = "emotion_transcript_utf8.txt"  # 入力ファイル名を固定化
#    file_path = "挨拶.txt"  # 入力ファイル名を固定化
    lines = read_file(file_path)
    ix = 0
# 
    dir_path = 'wav'
    os.makedirs(dir_path, exist_ok=True)  # 既に存在していてもエラーにならない

# 録音中かどうかを示すフラグ
    recording = False
    starting = False

# Tkinterウィンドウを作成
    root = tk.Tk()
    root.title("Speech Recording App")
    frame = tk.Frame(root)
    frame.pack(padx=20,pady=10,fill="x")

    listvar=tk.StringVar()
    listvar.set(lines)
    listbox=tk.Listbox(frame, width=120, height=3,  listvariable=listvar, selectmode='SINGLE')
    scrollbar = tk.Scrollbar(frame, orient=tk.VERTICAL, command=listbox.yview)
    listbox["yscrollcommand"] = scrollbar.set
    scrollbar.pack(side="right")
    listbox.pack(side="left",fill="x")
    listbox.selection_set(0) 

# 録音ボタンを作成
    record_button = tk.Button(root, text="Start Recording", command=toggle_recording) 
    record_button.pack(fill="x")

# 認識結果を表示するテキストウィジェットを作成
    output_text = tk_scrolledtext.ScrolledText(root, width=70, height=10, font=('Arial', 14))
    output_text.pack(pady=10)
    check_val = tk.BooleanVar()
    checkbutton = tk.Checkbutton(text="play", variable=check_val)
    checkbutton.pack()

    root.protocol("WM_DELETE_WINDOW", delete_window)
    # GUIを表示
    root.mainloop()