import argparse import os from loguru import logger def extract_list(folder_path, language, name, transcript_txt_file): logger.info(f"extracting list: {folder_path}|{name}|{language}") current_dir = os.getcwd() relative_path = os.path.relpath(folder_path, current_dir) print(relative_path) os.makedirs(os.path.dirname(transcript_txt_file), exist_ok=True) with open(transcript_txt_file, "w", encoding="utf-8") as f: # 遍历 raw 文件夹下的所有子文件夹 for root, _, files in os.walk(relative_path): for file in files: if file.endswith(".lab"): lab_file_path = os.path.join(root, file) # 读取转写文本 with open(lab_file_path, "r", encoding="utf-8") as lab_file: transcription = lab_file.read().strip() if len(transcription) == 0: continue # 获取对应的 WAV 文件路径 # ./Data/宵宫/audios/raw # ./Data/宵宫/audios/wavs wav_file_path = os.path.splitext(lab_file_path)[0] + ".wav" if os.path.isfile(wav_file_path): wav_file_path = wav_file_path.replace("\\", "/").replace( "/raw", "/wavs" ) # 写入数据到总的转写文本文件 line = f"{wav_file_path}|{name}|{language}|{transcription}\n" f.write(line) else: print("not exists!") return f"转写文本 {transcript_txt_file} 生成完成" if __name__ == "__main__": parser = argparse.ArgumentParser() parser.add_argument( "-f", "--filepath", required=True, help="path of your rawaudios, e.g. ./Data/xxx/audios/raw", ) parser.add_argument("-l", "--language", default="ZH", help="language") parser.add_argument("-n", "--name", required=True, help="name of the character") parser.add_argument("-o", "--outfile", required=True, help="outfile") args = parser.parse_args() status_str = extract_list(args.filepath, args.language, args.name, args.outfile) logger.critical(status_str)