python 通过麦克风录音 生成wav文件

#!/usr/bin/env python

# -*- coding: utf-8 -*-

########################################################################

# 

# Copyright (c) 2017 aibot.me, Inc. All Rights Reserved

# 

########################################################################

 

"""

File: gen_wav.py

Date: 2017/03/24 12:36:27

Brief: 通过麦克风录音 生成 wav文件

"""

import os

import sys

import wave

import numpy  as np 

from datetime  import datetime

from pyaudio  import PyAudio, paInt16

class  GenAudio( object):

     def  __init__( self):

         self.num_samples =  2000     #pyaudio内置缓冲大小

         self.sampling_rate =  8000   #取样频率

         self.level =  1500           #声音保存的阈值

         self.count_num =  20         #count_num个取样之内出现COUNT_NUM个大于LEVEL的取样则记录声音

         self.save_length =  8        #声音记录的最小长度：save_length * num_samples 个取样

         self.time_count =  8         #录音时间，单位s

         self.voice_string = []

    

     #保存文件

     def  save_wav( self,  filename):

        wf = wave.open(filename,  'wb') 

        wf.setnchannels( 1) 

        wf.setsampwidth( 2) 

        wf.setframerate( self.sampling_rate) 

        wf.writeframes(np.array( self.voice_string).tostring())

        wf.close()

    

    

     def  read_audio( self):

        pa = PyAudio() 

        stream = pa.open( format=paInt16,  channels= 1,  rate= self.sampling_rate,  input= True, 

                 frames_per_buffer= self.num_samples) 

        

        save_count =  0

        save_buffer = [] 

        time_count =  self.time_count

         while  True:

            time_count -=  1

            

             # 读入num_samples个取样

            string_audio_data = stream.read( self.num_samples)     

             # 将读入的数据转换为数组

            audio_data = np.fromstring(string_audio_data,  dtype = np.short)

             #计算大于 level 的取样的个数

            large_sample_count = np.sum(audio_data >  self.level)

            

             print(np.max(audio_data)),   "large_sample_count=>", large_sample_count

             # 如果个数大于COUNT_NUM，则至少保存SAVE_LENGTH个块

             if large_sample_count >  self.count_num:

                save_count =  self.save_length

             else: 

                save_count -=  1

             if save_count <  0:

                save_count =  0

            

             if save_count >  0:

                save_buffer.append(string_audio_data)

             else:

                 if  len(save_buffer) >  0:

                     self.voice_string = save_buffer

                    save_buffer = [] 

                     print( "Recode a piece of  voice successfully!")

                     return  True

            

             if time_count ==  0: 

                 if  len(save_buffer) >  0:

                     self.voice_string = save_buffer

                    save_buffer = []

                     print( "Recode a piece of  voice successfully!")

                     return  True

                 else:

                     return  False

         return  True

if  __name__ ==  "__main__":

    r = GenAudio()

    r.read_audio()

    r.save_wav( "./test.wav")