一个简单的英语音频分割

gistfile1.txt

#coding=utf-8
from pydub import AudioSegment
from pydub.silence import split_on_silence
import numpy as np
import matplotlib.pyplot as plt
import pyaudio
import os

file_folder = 'D:/Mirai/English/3L英语第二册/'



def Divide(file_name):
    id =  int(file_name.split('/')[-1].split('_')[-2])
    out_name = '3L_Unit_%.3d' % id
    sound = AudioSegment.from_mp3(file_name)
    data = np.fromstring(sound.raw_data,np.int16)
    #plt.plot(data)
    #plt.show()
    chunks = split_on_silence(sound, min_silence_len = 500, silence_thresh = -6000) # 小于-8000时无法分割
    '''
        p = pyaudio.PyAudio()

        stream = p.open(format = p.get_format_from_width(sound.sample_width),
                        channels = sound.channels,
                        rate = sound.frame_rate,
                        output = True)
    '''
           
    silence = AudioSegment.silent(duration=1500)             
    os.system('mkdir ' + out_name)   
    for i in range(1,len(chunks)):            
        #cdata = chunks[i].raw_data
        #stream.write(cdata)
        name = out_name + '//' + out_name + '_%.2d' %i + '.mp3'
        u = chunks[i] + silence
        u.export(name, format = 'mp3')
        #raw_input("next %d"% i)
    '''        
        stream.close()
        p.terminate()
    '''

for w in os.listdir(file_folder):
    Divide(file_folder + w)
    print 'ok',w