Python FFMPEG AttributeError : 'Popen' object has no attribute 'proc'

Question

我正在开发一个从音频流中学习的 tensorflow 项目。我正在尝试打开一个音频文件并使用 FFMPEG 将数据存储在一个数组中。我正在关注教程 here

我的代码如下所示:

import subprocess as sp
FFMPEG_BIN = "ffmpeg"
try:
    if image_file != 'train/rock/.DS_Store':
        command = [FFMPEG_BIN,
            '-i', image_file,
            '-f', 's16le',
            '-acodec', 'pcm_s16le',
            '-ar', '44100',
            '-ac', '2',
            'output.png']
        pipe = sp.Popen(command, stdout=sp.PIPE, bufsize=10**8)
        # pipe = sp.Popen(command, stdout=sp.PIPE)
        raw_audio = pipe.proc.stdout.read(88200*4)

但我得到了错误:

AttributeError: 'Popen' object has no attribute 'proc'

Answer 1

我正在使用 ffmpeg和 pyaudio .这段代码对我有用。

import pyaudio   
import subprocess as sp  
import numpy  
command = [ 'ffmpeg',
        '-i', "Filename", # I used a url stream 
    '-loglevel','error',
        '-f', 's16le',
        '-acodec', 'pcm_s16le',
        '-ar', '44100', # ouput will have 44100 Hz
        '-ac', '2', # stereo (set to '1' for mono)
        '-']  
pipe = sp.Popen(command, stdout=sp.PIPE, bufsize=10**8)  
p = pyaudio.PyAudio()  #PyAudio helps to reproduce raw data in pipe.  
stream = p.open(format = pyaudio.paInt16,
                channels = 2,
                rate = 44100,
                output = True)  
while True:  
    raw_audio = pipe.stdout.read(44100*2) #get raw data   
    stream.write(raw_audio) # reproduce  
    # Convert raw data in array with numpy  
    audio_array = numpy.fromstring(raw_audio, dtype="int16")  
    audio_array = audio_array.reshape((len(audio_array)/2,2))  
stream.stop_stream()  
stream.close()

在ubuntu中你可以安装 pyaudio和:

sudo apt-get install python-pyaudio python3-pyaudio 

或者

pip install pyaudio