2

Im trying to place a video on top of a background image but the output video is getting generated without audio. Is there any way to keep the audio as it is?

def ConvertVideo(source,background,start,end,dest):
    stream = ffmpeg.input(source)
    strea1 = ffmpeg.input(background)
    duration = end - start
    stream = stream.trim(start=start,duration=duration).filter('setpts', 'PTS-STARTPTS')
    stream = stream.crop(100,0,1080,720)
    stream = ffmpeg.overlay(strea1,stream,x=0,y=180)
    stream = stream.output(dest)

Does anyone know why audio gets dropped? Is there any workaround to this problem?

oneclick
  • 91
  • 2
  • 9
  • what does `ffmpeg.compile(stream)` return? – kesh Feb 07 '22 at 19:22
  • Where have you found documentation about the 'stream.trim()'' function? I can't find it anywhere! How come it is missing even from ffmpeg.org itself? – Apostolos Jul 21 '23 at 22:50

1 Answers1

1

We have to add the audio stream explicitly.
Something like: ffmpeg.output(overlaid_vid_stream, audio_stream, dest).

Start by creating synthetic input video and background using FFmpeg CLI (just for making the posted solution reproducible and "self contained").

  • Create input video with audio:

     ffmpeg -y -f lavfi -i testsrc=size=128x72:rate=1 -f lavfi -i sine=frequency=400 -f lavfi -i sine=frequency=1000 -filter_complex "[1:a][2:a]amix=inputs=2" -vcodec libx264 -g 10 -crf 17 -pix_fmt yuv420p -acodec aac -ar 22050 -t 50 in.mp4

  • Create yellow background image:

     ffmpeg.exe -y -f lavfi -i color=yellow:size=128x72 -frames:v 1 background.png

For the example, the size is 128x72 (assuming your input is 1280x720).

The following code sample applies trim, setpts, crop and overlay filters, and adds the source audio:

source = 'in.mp4'
background = 'background.png'
dest = 'out.mp4'
end = 40
start = 10

duration = end - start

vid_stream = ffmpeg.input(source).video  # Source video stream
audio_stream = ffmpeg.input(source).audio  # Source audio stream
vid_background = ffmpeg.input(background).video  # Background video stream

trimed_vid_stream = vid_stream.trim(start=start, duration=duration).filter('setpts', 'PTS-STARTPTS').crop(10, 0, 108, 72)  # Source video stream after trimming and cropping
overlaid_vid_stream = ffmpeg.overlay(vid_background, trimed_vid_stream, x=0, y=18)  # Video stream overlay of background and trimed_vid_stream

trimed_audio_stream = audio_stream.filter('atrim', start=start, duration=duration).filter('asetpts', 'PTS-STARTPTS')  # Trimming the audio
output_video_and_audio = ffmpeg.output(overlaid_vid_stream, trimed_audio_stream, dest)  # Output - video applies overlaid_vid_stream, and audio applies trimmed source audio

output_video_and_audio.overwrite_output().run() # Execute FFmpeg

I modified your naming to be more meaningful (naming everything stream makes it difficult to follow).
I also removed the method, and set arguments to specific values (just for making the posted solution reproducible).

The solution applies the following main stages:

  • Create references to the video and the audio streams:

     vid_stream = ffmpeg.input(source).video  # Source video stream
 audio_stream = ffmpeg.input(source).audio  # Source audio stream

  • Define video filters:

     vid_background = ffmpeg.input(background).video  # Background video stream
 trimed_vid_stream = vid_stream.trim(start=start, duration=duration).filter('setpts', 'PTS-STARTPTS').crop(10, 0, 108, 72)  # Source video stream after trimming and cropping
 overlaid_vid_stream = ffmpeg.overlay(vid_background, trimed_vid_stream, x=0, y=18)  # Video stream overlay of background and trimed_vid_stream

  • Define audio filters (trimming the audio):

     trimed_audio_stream = audio_stream.filter('atrim', start=start, duration=duration).filter('asetpts', 'PTS-STARTPTS')  # Trimming the audio

  • Define the output to include the "filtered" video and the trimmed audio:

     output_video_and_audio = ffmpeg.output(overlaid_vid_stream, trimed_audio_stream, dest)  # Output - video applies overlaid_vid_stream, and audio applies trimmed source audio

  • Execute FFmpeg:

     output_video_and_audio.overwrite_output().run() # Execute FFmpeg

Sample output frame:
enter image description here

The video includes a beeping audio...

Rotem
  • 30,366
  • 4
  • 32
  • 65
  • this is great buddy but how to trim the audio according to the video output... i tried using ```audio.filter('atrim', duration=5)``` this kind of didn't worked – oneclick Feb 08 '22 at 06:17
  • I updated my answer... Since my sample audio is just a sine wave, I was not aware of the the audio issue. Syntax for trimming the audio: `trimed_audio_stream = audio_stream.filter('atrim', start=start, duration=duration).filter('asetpts', 'PTS-STARTPTS')` . – Rotem Feb 08 '22 at 09:20
  • hi. can anyone tell me how can i convert compress a video using ffmpeg-python code without loss in its quality and how can i convert a video of 1080p to 720p, 480p and 360p can you help me with these two problems – sandeepnegi Aug 30 '22 at 07:24
  • @sandeepnegi I can't answer you by comment... I suggest you to look for solution using FFmpeg in command line first. (In case you decide to post a question, I am probably not going to answer it, but someone else may). – Rotem Aug 30 '22 at 07:41
  • i have done it . for others reference they can use **filter** with **scale** and provide height and width of the video like this ```video_stream = video_stream.filter('scale', w=1280, h=720)``` . – sandeepnegi Aug 30 '22 at 12:16