博客

"""
把原音频，按csv中的标注结果分割成几个音频，如text是无，忽略。否则从sDate到eDate 进行分割。
"""
import os
import pandas as pd
import json
from pydub import AudioSegment

csv_url = "D:\csv\\"
wav_url = "D:\标注音频与示例\\"
save_wav = "D:\wav\\new_wav\\"

for path in os.listdir(csv_url):
    data_frame = pd.read_csv(csv_url + path, encoding='utf-8')
    name_list = data_frame["storeFileName"]
    result_list = data_frame["finalResult"]
    for nl, rl in zip(name_list, result_list):
        new_wav_url = save_wav + nl.split(".")[0] + "\\"
        if not os.path.exists(new_wav_url):
            os.makedirs(new_wav_url)
        result = json.loads(rl)
        txt_url = new_wav_url + nl.split(".")[0]+".txt"
        with open(txt_url, "w", encoding='utf-8') as fn:
            fn.write(nl+"\n")
            fn.write(rl)
            fn.close()
        audio = AudioSegment.from_wav(wav_url + nl)
        for text in result['text']:
            if text['defData']['text'] != '无':
                sTime = text['defData']['sTime'] * 1000
                dTime = text['defData']['dTime'] * 1000
                eTime = sTime + dTime
                # 音频切割按开始时间到结束时间切割
                audio_chunk = audio[sTime:eTime]

                audio_chunk.export(new_wav_url + nl.split(".")[0] + "-" + str(text['id']) + ".wav", format="wav")