属性错误：找不到 PyAudio；检查安装...无法使用语音识别

Question

我正在尝试制作一个基本的语音识别助手。当我运行代码时，它告诉我：

Traceback (most recent call last):
  File "C:\Users\Mo.haytham\AppData\Local\Programs\Python\Python37\lib\site-packages\speech_recognition\__init__.py", line 108, in get_pyaudio
    import pyaudio
ModuleNotFoundError: No module named 'pyaudio'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "C:/Users/Mo.haytham/.PyCharmCE2018.3/config/scratches/ALPHA_BASIC.py", line 22, in <module>
    hear()
  File "C:/Users/Mo.haytham/.PyCharmCE2018.3/config/scratches/ALPHA_BASIC.py", line 13, in hear
    with sr.Microphone() as sourse:
  File "C:\Users\Mo.haytham\AppData\Local\Programs\Python\Python37\lib\site-packages\speech_recognition\__init__.py", line 79, in __init__
    self.pyaudio_module = self.get_pyaudio()
  File "C:\Users\Mo.haytham\AppData\Local\Programs\Python\Python37\lib\site-packages\speech_recognition\__init__.py", line 110, in get_pyaudio
    raise AttributeError("Could not find PyAudio; check installation")
AttributeError: Could not find PyAudio; check installation 

我尝试过，pip install pyaudio但随后出现此错误：

Running setup.py clean for pyaudio
Failed to build pyaudio
Installing collected packages: pyaudio
  Running setup.py install for pyaudio ... error
    ERROR: Complete output from command 'c:\users\mo.haytham\appdata\local\programs\python\python37\python.exe' -u -c 'import setuptools, tokenize;__file__='"'"'C:\\Users\\MO2D8C~1.HAY\\AppData\\Local\\Temp\\pip-install-o2
10x3zl\\pyaudio\\setup.py'"'"';f=getattr(tokenize, '"'"'open'"'"', open)(__file__);code=f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, __file__, '"'"'exec'"'"'))' install --record 'C:\Users\MO2
D8C~1.HAY\AppData\Local\Temp\pip-record-hr7kket1\install-record.txt' --single-version-externally-managed --compile:
    ERROR: running install
    running build
    running build_py
    creating build
    creating build\lib.win-amd64-3.7
    copying src\pyaudio.py -> build\lib.win-amd64-3.7
    running build_ext
    building '_portaudio' extension
    error: Microsoft Visual C++ 14.0 is required. Get it with "Microsoft Visual C++ Build Tools": https://visualstudio.microsoft.com/downloads/
    ----------------------------------------
ERROR: Command "'c:\users\mo.haytham\appdata\local\programs\python\python37\python.exe' -u -c 'import setuptools, tokenize;__file__='"'"'C:\\Users\\MO2D8C~1.HAY\\AppData\\Local\\Temp\\pip-install-o210x3zl\\pyaudio\\setup.p
y'"'"';f=getattr(tokenize, '"'"'open'"'"', open)(__file__);code=f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, __file__, '"'"'exec'"'"'))' install --record 'C:\Users\MO2D8C~1.HAY\AppData\Local\
Temp\pip-record-hr7kket1\install-record.txt' --single-version-externally-managed --compile" failed with error code 1 in C:\Users\MO2D8C~1.HAY\AppData\Local\Temp\pip-install-o210x3zl\pyaudio\

def hear():
    import speech_recognition as sr
    ear = sr.Recognizer()
    with sr.Microphone() as sourse:
        print("listening...")
        audio = ear.listen(sourse)
        try:
            text = ear.recognize_google(audio)
            print(text)
        except:
            print("i didn't get that...")

hear()

Answer 1

在终端类型

pip install pipwin

然后

pipwin install pyaudio

Answer 2

我还发现 PyAudio 安装可能会很痛苦，因为安装困难，甚至对某些最终用户来说也是个难题。原则上没有理由为什么speech_recognition.Recognizer.listen()不能从其他音频库，如采取其输入sounddevice或声卡或audiomath，所有这些都是通常更容易安装。幸运的是，虽然speech_recognition代码本身只提供了 PyAudio 实现，但在内部它只需要Microphone鸭子类型的几个属性才能使其listen()成功。具体来说：

• source必须是speech_recognition.AudioSource子类的实例
• source.streamNone当源处于活动状态时必须是非
• source.CHUNK 必须是每个块的（整数）样本数
• source.SAMPLE_RATE 必须是采样率
• source.SAMPLE_WIDTH 必须是每个样本的字节数
• source.stream.read(numberOfSamples) 必须返回原始单声道音频数据

这是使用audiomath的鸭式解决方案：

import audiomath; audiomath.RequireAudiomathVersion( '1.12.0' )
import speech_recognition  # NB: python -m pip install SpeechRecognition

class DuckTypedMicrophone( speech_recognition.AudioSource ): # descent from AudioSource is required purely to pass an assertion in Recognizer.listen()
    def __init__( self, device=None, chunkSeconds=1024/44100.0 ):  # 1024 samples at 44100 Hz is about 23 ms
        self.recorder = None
        self.device = device
        self.chunkSeconds = chunkSeconds
    def __enter__( self ):
        self.nSamplesRead = 0
        self.recorder = audiomath.Recorder( audiomath.Sound( 5, nChannels=1 ), loop=True, device=self.device )
        # Attributes required by Recognizer.listen():
        self.CHUNK = audiomath.SecondsToSamples( self.chunkSeconds, self.recorder.fs, int )
        self.SAMPLE_RATE = int( self.recorder.fs )
        self.SAMPLE_WIDTH = self.recorder.sound.nbytes
        return self
    def __exit__( self, *blx ):
        self.recorder.Stop()
        self.recorder = None
    def read( self, nSamples ):
        sampleArray = self.recorder.ReadSamples( self.nSamplesRead, nSamples )
        self.nSamplesRead += nSamples
        return self.recorder.sound.dat2str( sampleArray )
    @property
    def stream( self ): # attribute must be present to pass an assertion in Recognizer.listen(), and its value must have a .read() method
        return self if self.recorder else None

if __name__ == '__main__':
    import speech_recognition as sr
    r = sr.Recognizer()
    with DuckTypedMicrophone() as source:
        print('\nSay something to the %s...' % source.__class__.__name__)
        audio = r.listen(source)
    print('Got it.')
    print('\nUnderstood: "%s"\n' % r.recognize_google(audio))

    if 0: # plot and/or play back captured audio
        s = audiomath.Sound(audio.get_wav_data(), fs=audio.sample_rate, nChannels=1)
        s.Play()
        s.Plot()

Answer 3

如果您是 ubuntu 18.04 用户，请按照以下步骤操作

sudo apt-get install portaudio19-dev python-pyaudio

然后

pip install PyAudio