python利用不到一百行代码实现一个小siri

# coding=utf8

importos

importwave

importdtw

importnumpy as np

importpyaudio

defcompute_distance_vec(vec1, vec2):

 returnnp.linalg.norm(vec1-vec2)#计算两个特征之间的欧氏距离

classrecord():

 defrecord(self, CHUNK=44100,FORMAT=pyaudio.paInt16, CHANNELS=2, RATE=44100, RECORD_SECONDS=200,

    WAVE_OUTPUT_FILENAME="record.wav"):

  #录歌方法

  p=pyaudio.PyAudio()

  stream=p.open(format=FORMAT,

      channels=CHANNELS,

      rate=RATE,

      input=True,

      frames_per_buffer=CHUNK)

  frames=[]

  foriinrange(0,int(RATE/CHUNK*RECORD_SECONDS)):

   data=stream.read(CHUNK)

   frames.append(data)

  stream.stop_stream()

  stream.close()

  p.terminate()

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

  wf.setnchannels(CHANNELS)

  wf.setsampwidth(p.get_sample_size(FORMAT))

  wf.setframerate(RATE)

  wf.writeframes(''.join(frames))

  wf.close()

classvoice():

 defloaddata(self, filepath):

  try:

   f=wave.open(filepath,'rb')

   params=f.getparams()

   self.nchannels,self.sampwidth,self.framerate,self.nframes=params[:4]

   str_data=f.readframes(self.nframes)

   self.wave_data=np.fromstring(str_data, dtype=np.short)

   self.wave_data.shape=-1,self.sampwidth

   self.wave_data=self.wave_data.T#存储歌曲原始数组

   f.close()

   self.name=os.path.basename(filepath)# 记录下文件名

   returnTrue

  except:

   raiseIOError,'File Error'

 deffft(self, frames=40):

  self.fft_blocks=[]#将音频每秒分成40块，再对每块做傅里叶变换

  blocks_size=self.framerate/frames

  foriinxrange(0,len(self.wave_data[0])-blocks_size, blocks_size):

   self.fft_blocks.append(np.abs(np.fft.fft(self.wave_data[0][i:i+blocks_size])))

 @staticmethod

 defplay(filepath):

  chunk=1024

  wf=wave.open(filepath,'rb')

  p=pyaudio.PyAudio()

  # 播放音乐方法

  stream=p.open(format=p.get_format_from_width(wf.getsampwidth()),

      channels=wf.getnchannels(),

      rate=wf.getframerate(),

      output=True)

  whileTrue:

   data=wf.readframes(chunk)

   ifdata=="":break

   stream.write(data)

  stream.close()

  p.terminate()

if__name__=='__main__':

 r=record()

 r.record(RECORD_SECONDS=3, WAVE_OUTPUT_FILENAME='record.wav')

 v=voice()

 v.loaddata('record.wav')

 v.fft()

 file_list=os.listdir(os.getcwd())

 res=[]

 foriinfile_list:

  ifi.split('.')[1]=='wav'andi.split('.')[0] !='record':

   temp=voice()

   temp.loaddata(i)

   temp.fft()

   res.append((dtw.dtw(v.fft_blocks, temp.fft_blocks, compute_distance_vec)[0],i))

 res.sort()

 printres

 ifres[0][1].find('open_qq') !=-1:

  os.system('C:\program\Tencent\QQ\Bin\QQScLauncher.exe')#我的QQ路径

 elifres[0][1].find('zhimakaimen') !=-1:

  os.system('chrome.exe')#浏览器的路径，之前已经被添加到了Path中了

 elifres[0][1].find('play_music') !=-1:

  voice.play('C:\data\music\\audio\\audio\\ (9).wav')#播放一段音乐

 # r = record()

 # r.record(RECORD_SECONDS=3,WAVE_OUTPUT_FILENAME='zhimakaimen_09.wav')