在 Tkinter 帧上播放带音频的视频

Question

所以我正在尝试做一些非常笨拙的事情。我正在使用 opencv 和 ffpyplayer 尝试在 Tk 窗口上制作嵌入式视频播放器。

它确实有效，但我在同步音频和视频时遇到问题，我可以接近一个不错的结果，但过了一会儿它又不同步了。

完整代码：

import time, traceback, os, telepot
from tkinter import *
import cv2, youtube_dl # pip install opencv-python; pip install --upgrade 
youtube_dl
from PIL import Image, ImageTk #
from io import BytesIO # io
from ffpyplayer.player import MediaPlayer # pip install ffpyplayer
from pytube import YouTube

_name_ = os.path.basename(os.path.realpath(__file__))
_path_ = os.path.realpath(__file__).replace(_name_, '')

class Screen(Frame):
    '''
        Screen widget: Embedded video player from local or youtube
    '''
    def __init__(self, parent, *args, **kwargs):
        Frame.__init__(self, parent, bg = 'black')
        self.settings = { # Inizialazing dictionary settings
            "width" : 1024,
            "height" : 576
        }
        self.settings.update(kwargs) # Changing the default settings
        # Open the video source |temporary
        self.video_source =  _path_+'asd.mp4'

        # Inizializing video and audio variables
        self.vid = None
        self.aux = None
        # Canvas of the player
        self.canvas = Canvas(self, width = self.settings['width'], height = self.settings['height'], bg = "black", highlightthickness = 0)
        self.canvas.pack()

        # NEED TO SYNC AUDIO
        self.delay = 15 # Delay between frames of player

    def update(self):
        '''
            Function: Start the player and keeps drawing the canvas 
        '''
        if not self.vid or not self.aux: # If Audio or Video is missing stop everything
            self.stop()
            return None

        # Get the frames and if video and audio are running
        ret, frame = self.get_frame()
        audio_frame, val = self.aux.get_frame()

        # Drawing frames on canvas
        if self.fb == 1: # Check if it's the first cycle, trying to make the audio start with the video
            self.photo = ImageTk.PhotoImage(image = Image.fromarray(frame).resize((self.settings['width'], self.settings['height'])))
            self.canvas.create_image(0,0, image = self.photo, anchor = 'nw')
            self.fb = 0
            self.aux.set_pause(False) # Starting the audio
        elif ret and val != 'eof':
            self.photo = ImageTk.PhotoImage(image = Image.fromarray(frame).resize((self.settings['width'], self.settings['height'])))
            self.canvas.create_image(0,0, image = self.photo, anchor = 'nw')

        self.after(self.delay, self.update) # Update for single frame, need to sync

    def get_frame(self):
        '''
            Function: Draws the frames
        '''
        if self.vid.isOpened():
            ret, frame = self.vid.read()
            if ret:
                # Return a boolean success flag and the current frame converted to BGR
                return (ret, cv2.cvtColor(frame, cv2.COLOR_BGR2RGB))
            else:
                return (ret, None)

    def youTube(self, ID):
        '''
            Function: Gets the youtube video and starts it 
        '''
        print("(TO REPLACE) : Downloading")
        yt = YouTube("https://www.youtube.com/watch?v=" + ID)
        stream = yt.streams.filter(progressive=True).first() # SEE THE POSSIBLE THINGS TO DOWNLOAD
        stream.download(_path_, 'test')
        print("(TO REPLACE) : Finished")
        self.start(_path_+'\\test.mp4')

    def start(self, _source):
        '''
            Function: Starts the player when gets input from keyboard(temporal) or Telegram
        '''
        try: # Stopping player if is already playing for a new video
            self.stop()
        except:
            None

        ff_opts = {'paused' : True} # Audio options
        self.fb = 1 # Setting first cycle

        if _source == 'local': # Checking which source use
            self.vid = cv2.VideoCapture(self.video_source)
            self.aux = MediaPlayer(self.video_source, ff_opts=ff_opts)
        else:
            self.vid = cv2.VideoCapture(_source)
            self.aux = MediaPlayer(_source, ff_opts=ff_opts)

        if not self.vid.isOpened():
            raise ValueError("Unable to open video source")

        self.update() # Starting the player

    def stop(self):
        '''
            Function: Release and stop Video and Audio
        '''
        try: # Stopping video
            self.vid.release()
            self.vid = None
        except:
            pass
        try: # Stopping audio
            self.aux.toggle_pause()
            self.aux = None
        except:
            pass
        self.canvas.delete('all') # Resetting canvas

    def __del__(self):
        '''
            Function: Release the video source when the object is destroyed
        '''
        if self.vid.isOpened():
            self.vid.release()

class Mirror:
    '''
        Mainframe: Display where to put the widgets
    '''
    def __init__(self):
        self.tk = Tk() # Creating the window
        self.tk.configure(background = 'black')
        self.tk.update()

        # Setting up the FRAMES for widgets
        self.bottomFrame = Frame(self.tk, background = 'black')
        self.bottomFrame.pack(side = BOTTOM, fill = BOTH, expand = YES)

        # Bindings and fullscreen setting
        self.fullscreen = False
        self.tk.bind("<Return>", self.toggle_fullscreen)
        self.tk.bind("<Escape>", self.end_fullscreen)

        # Screen, BOT
        print("Inizializing Screen...")
        self.screen = Screen(self.bottomFrame)
        self.screen.pack(side = TOP)

        self.tk.bind("<Key>", self.key) # Get inputs from keyboard

    def key(self, event):
        pressed = repr(event.char).replace("'", '')
        if pressed == 's':
            self.screen.stop()
        elif pressed == 'a':
            self.screen.start('local')
        else:
            print('fail')

    def toggle_fullscreen(self, event = None):
        self.fullscreen = True
        self.tk.attributes("-fullscreen", self.fullscreen)

    def end_fullscreen(self, event = None):
        self.fullscreen = False
        self.tk.attributes("-fullscreen", self.fullscreen)

    def on_chat_message(msg):
        content_type, chat_type, chat_id = telepot.glance(msg)
        message = str(msg.get('text'))

        if 'https://youtu.be/' in message:
            URL_VIDEO = message.split('https://youtu.be/')[1]
            Mir.screen.youTube(URL_VIDEO)       
        elif 'stop' == message.lower():
            Mir.screen.stop()

if __name__ == '__main__':
    Mir = Mirror()
    #bot = telepot.Bot(TELEGRAM_TOKEN)
    #bot.message_loop(on_chat_message)
    Mir.tk.mainloop()
    #while 1:
        #time.sleep(10)

具体渲染帧的方法：

# NEED TO SYNC AUDIO
    self.delay = 15 # Delay between frames of player

def update(self):
    '''
        Function: Start the player and keeps drawing the canvas 
    '''
    if not self.vid or not self.aux: # If Audio or Video is missing stop everything
        self.stop()
        return None

    # Get the frames and if video and audio are running
    ret, frame = self.get_frame()
    audio_frame, val = self.aux.get_frame()

    # Drawing frames on canvas
    if self.fb == 1: # Check if it's the first cycle, trying to make the audio start with the video
        self.photo = ImageTk.PhotoImage(image = Image.fromarray(frame).resize((self.settings['width'], self.settings['height'])))
        self.canvas.create_image(0,0, image = self.photo, anchor = 'nw')
        self.fb = 0
        self.aux.set_pause(False) # Starting the audio
    elif ret and val != 'eof':
        self.photo = ImageTk.PhotoImage(image = Image.fromarray(frame).resize((self.settings['width'], self.settings['height'])))
        self.canvas.create_image(0,0, image = self.photo, anchor = 'nw')

    self.after(self.delay, self.update) # Update for single frame, need to sync

我尝试了其他方法来使这件事发挥作用，并查看是否可以找到解决方案但没有成功，如果有人有一些想法或更好的解决方案来做到这一点，我将不胜感激。

我还尝试从 youtube 获取视频作为流而不需要先下载整个视频来播放它，我可以让视频正常工作，但我找不到让音频播放的方法，如果有人知道一种方法，我也想知道。

编辑： 所以视频的帧速率应该是 24、30 或 60，我应该检查视频的帧速率然后相应地设置延迟，我这样做的方式是我根据视频手动更改延迟帧速率试图通过反复试验使其同步。老实说，我对音频知之甚少，所以对那些东西很不屑一顾..

为了使整个代码运行，它需要来自电报机器人的令牌或本地视频文件才能播放。

Answer 1

好吧，我找到了一个更好的解决方案。我尝试再次使用 vlc 模块，之前我在使用它时遇到了问题，但在我完成它之后，我实际上找到了一种将 vlc 的输出打印到 带有画布的框架的方法。

其实很简单，我去掉了cv2和ffpyplayer，用了一个简单的VLC播放器

请注意，在我的情况下，屏幕是一个框架，所以要完成这项工作，只需使用 'Frame'.winfo_id() 来获取输出视频、音频的 tk 框架的 id也可以。然后只需使用播放器 .set_hwnd(ID) 将其设置为 vlc 播放器。

class Screen(Frame):
'''
    Screen widget: Embedded video player from local or youtube
'''
def __init__(self, parent, *args, **kwargs):
    Frame.__init__(self, parent, bg = 'black')
    self.settings = { # Inizialazing dictionary settings
        "width" : 1024,
        "height" : 576
    }
    self.settings.update(kwargs) # Changing the default settings
    # Open the video source |temporary
    self.video_source =  _path_+'asd.mp4'

    # Canvas where to draw video output
    self.canvas = Canvas(self, width = self.settings['width'], height = self.settings['height'], bg = "black", highlightthickness = 0)
    self.canvas.pack()

    # Creating VLC player
    self.instance = vlc.Instance()
    self.player = self.instance.media_player_new()


def GetHandle(self):
    # Getting frame ID
    return self.winfo_id()

def play(self, _source):
    # Function to start player from given source
    Media = self.instance.media_new(_source)
    Media.get_mrl()
    self.player.set_media(Media)

    #self.player.play()
    self.player.set_hwnd(self.GetHandle())
    self.player.play()