Recherche avancée

Sur d’autres sites (9056)

• FFMPEG : Remove silence with exact INTERVALS detected by silencedetect Filter

  29 mai 2021, par AfricanMamba

  I am trying to remove silence given the specific intervals given the output of the silencedetect filter. I recently read this post, but in my case, I want to use ONLY the intervals produced by the filter while using the same noise threshold. Here is what I have tried and the issues I am running into.

  


  After running the ffprobe :

  


  Input #0, wav, from 'audio.wav':
  Metadata:
    encoder         : Lavf58.76.100
  Duration: 00:00:56.21, bitrate: 1536 kb/s


  


  Here you can see that the video is exactly 56.21 seconds.

  


  After running silencedetect it produces an interval of silence with a threshold of -20dB :

  


  [silencedetect @ 000001dd38cead80] silence_start: 0
[silencedetect @ 000001dd38cead80] silence_end: 4.5631 | silence_duration: 4.5631
[silencedetect @ 000001dd38cead80] silence_start: 13.320
[silencedetect @ 000001dd38cead80] silence_end: 17.5232 | silence_duration: 4.2032


  


  After running this, I expected after using silenceremove, it would remove 8.7663 seconds of audio and the video would trim down to about 47.4437 seconds.

  


  The command silenceremove :

  


  ffmpeg -i audio.wav -af silenceremove=stop_periods=-1:stop_threshold=-20dB:window=0 output.wav


  


  After listening to the output I noticed that it cut out different regions of the audio that were not a part of the two intervals I was expecting it to remove. I ran the output in fprobe again to see the results :

  


  Input #0, wav, from 'output.wav':
  Metadata:
    encoder         : Lavf58.76.100
  Duration: 00:00:33.48, bitrate: 1536 kb/s


  


  It removed much more than the expected 8 seconds. Does anyone happen to know the command input parameters I need to use in order for it to only remove the intervals that it detected in silencedetect ? I do not think using duration will be the best tool because 1 : the two durations are different 2 : I am trying to create a program that removes the segments of silence automatically without me having to hard code it. Does anyone happen to know how I can solve this issue ? All help is appreciated !

  


• Adding a filter and title cards to user uploaded videos [on hold]

  5 septembre 2013, par Django Reinhardt

  A client is interested in applying a Sepia style filter, as well as title cards, to user-uploaded videos on their website. This would obviously be automated on their webserver, and it seems ffmpeg is a good way to make this happen.

  Unfortunately, from what I've read, such processing would require a lot of webserver CPU power — far more than the average webserver usually has. How can I practically perform these calculations on a website ?

• How to encode frames selectively and write in different video files as per required

  10 juin 2018, par MSD Paul

  suppose, for example, I have 10 frames, I want to encode 1,2,3 in one video, 4,5,6 in another video and rest in another video. How to achieve this without encoding each of them differently by us. (as that solution will not be scalable) How to change in the code while writing encoded frames into the video file ?

  How to make the write to happen into multiple video files.

  and how to change the sequence of encoding i.e. after 1 how can I possibly encode 4,5 skipping 2,3 after 4,5 done then encode 2,3. How to do this ?