vv_txt_wav_to_srt.py

import argparse
from datetime import datetime
import os
import subprocess
import sys
import yaml


def create_duration_list(wav_dir):
    if not wav_dir or not os.path.isdir(wav_dir):
        print(f'Does not exist directory. : {wav_dir}')
        sys.exit(1)

    duration_list = []
    for root, _, files in os.walk(wav_dir):
        files = sorted(files)  # ファイルをソート
        for file in files:
            if file.endswith(".wav"):
                filepath = os.path.join(root, file)
                result = subprocess.run(
                    ["ffprobe", "-i", filepath, "-show_entries", "format=duration", "-v", "quiet", "-of", "csv=p=0"],
                    stdout=subprocess.PIPE,
                    stderr=subprocess.PIPE,
                    text=True
                )
                duration = result.stdout.strip()
                duration_list.append((filepath, float(duration)))
    return duration_list


def create_text_list(text_file):
    text_list = []
    with open(text_file, "r") as f:
        for line in f:
            talk = line.strip().split(",", 1)[1]  # 最初の "," で分割し、右側の部分だけ取得
            text_list.append(talk)
    return text_list


def parse_time_to_seconds(time_str):
    try:
        dt = datetime.strptime(time_str, "%H:%M:%S,%f")  # ミリ秒あり
    except ValueError:
        dt = datetime.strptime(time_str, "%H:%M:%S")  # ミリ秒なし

    return dt.hour * 3600 + dt.minute * 60 + dt.second + dt.microsecond / 1_000_000


def read_setting_yaml(yaml_path):
    time_adjustments = {}
    if not yaml_path or not os.path.exists(yaml_path):
        return time_adjustments

    with open(yaml_path, "r") as f:
        setting = yaml.safe_load(f)
        for adjust in setting["time_adjustments"]:
            if "line" in adjust:
                seconds = parse_time_to_seconds(adjust["start_time"])
                time_adjustments[int(adjust["line"])] = seconds
    return time_adjustments


def convert_time(seconds):
    milliseconds = int((seconds - int(seconds)) * 1000)
    seconds = int(seconds)
    minutes, seconds = divmod(seconds, 60)
    hours, minutes = divmod(minutes, 60)
    return f"{hours:02}:{minutes:02}:{seconds:02},{milliseconds:03}"


def create_srt(output_srt_file, text_list, duration_list,time_adjustments):
    with open(output_srt_file, "w", encoding='utf-8') as f:
        index = 1
        start_time = 0
        end_time = 0
        for text, duration in zip(text_list, duration_list):
            start_time = end_time
            if index in time_adjustments:
                start_time = time_adjustments[index]    # 調整時間を適用
            end_time = start_time + duration[1]
            f.write(f"{index}\n")
            f.write(f"{convert_time(start_time)} --> {convert_time(end_time)}\n")
            f.write(f"{text}\n")
            f.write("\n")
            index += 1


def main():
    # コマンドライン引数のパーサーを設定
    parser = argparse.ArgumentParser(description="Process VOICEVOX text and .wav files to output srt file.")
    parser.add_argument("text_file", type=str, help="VOICEVOX text file")
    parser.add_argument("wav_dir", type=str, help="VOICEVOX .wav files directory")
    parser.add_argument("output_srt_file", type=str, help="output srt file")
    parser.add_argument('-s', '--setting', help='setting.yaml')
    args = parser.parse_args()

    text_list = create_text_list(args.text_file)
    duration_list = create_duration_list(args.wav_dir)
    time_adjustments = read_setting_yaml(args.setting)
    create_srt(args.output_srt_file, text_list, duration_list, time_adjustments)


if __name__ == "__main__":
    main()