data_preprocessing.py 文件源码

def convert_mp4(video_dir, audio_dir):
    '''
    Args: 
        1. video_dir:   Directory for all video files
        2. audio_dir:   Directory where all converted files will be stored.
    '''

    # Get all file names
    video_file_names = sorted(glob.glob(video_dir + "*.mp4"))
    # Extract actual names of file, also remove any extensions
    video_names = map(lambda x : x.split('/')[-1].split(".")[0], video_file_names)

    # Command for converting video to audio
    command = "ffmpeg -i " + video_dir + "{0}.mp4 -ab 96k -ar 44100 -vn " + audio_dir + "{0}.wav"

    for name in video_names:
        subprocess.call(command.format(name), shell=True)