Recherche avancée

Sur d’autres sites (17779)

• How do I use ffmpeg's `filter_complex` `volume` filter on videos that might not have an audio track ?

  9 mars 2023, par jaygooby

  I'm mixing new audio tracks into videos, and I want the original video audio to be at a lower volume in the mix :

  


  ffmpeg -i audio.wav -i video.mp4 -c:v copy -filter_complex "[1:a:0]volume=0.5[originalAudio]; [0:a:0][originalAudio]amix=inputs=2[m]" -map [m] -map 1:v:0 tmp.mp4


  


  that sets the original video audio to be 50% less than it was, when audio.wav is mixed with it. This works fine until there's a video with no audio track. Then I get the error :

  


  Stream specifier ':a:0' in filtergraph description [1:a:0]volume=0.05[originalAudio]; [0:a:0][originalAudio]amix=inputs=2[m] matches no streams.


  


  because the video has no audio track.

  


  Is there a way for me to use ffmpeg filters in the same call, to ensure there's an audio track if one is originally missing, so I don't need to check for the presence of an audio track using an additional step with ffprobe, mediainfo etc ?

  


  [Edit 1] Looks like I need to add a anullsrc track to the video inside the filter_complex, and then not specifically choose which audio track from the video I want, so either the original or the new silent one (if it's the only one present) is selected via ffmpeg's automatic stream selection...

  


  [Edit 2] Something like this, where I have 2 audio inputs, the first is silence and the second is the input wav. Use concat to pick either the silence or the video audio, use the result of that to reduce the volume, then mix in the new audio to the result of that, but I'm not there yet :

  


  ffmpeg -f lavfi -i anullsrc \
       -i audio.wav \
       -i video.mp4 \
       -c:v copy \
       -filter_complex \
       "[0:a] [2:a] concat=n=2:v=0:a=1 [silence-or-video-audio];
        [silence-or-video-audio]volume=0.5[reduced-audio]; [1:a:0][reduced-audio]amix=inputs=2[mixed-audio]" \
       -map "[mixed-audio]" \
       -map "[2:v:0]" \
       tmp.mp4


  


  The error is : Stream specifier ':a' in filtergraph description [0:a] [2:a] concat=n=2:v=0:a=1 [silence-or-video-audio]; [silence-or-video-audio]volume=0.5[reduced-audio]; [1:a:0][reduced-audio]amix=inputs=2[mixed-audio] matches no streams.

  


• options : handle options with the same name in codecs and formats.

  26 juillet 2011, par Clément Bœsch

  options : handle options with the same name in codecs and formats.

• ffmpeg to lower/fade audio volume of one audio stream when microphone voice detected ?

  11 juin 2021, par Lectos Lacious

  I want to do live audio translation via microphone, to get streamed live vid/audio from Facebook, plug the mic into laptop and do live translation by mixing existing audio stream with one coming from the mic (translation). This is OK, somehow I got this part by using audio filter "amix" and mix two audio streams together into one. Now I want to add more perfection to it, is it possible to (probably is) upon mic voice detection to automatically decrease/fade down 20% volume of input/original audio stream to hear translation (mic audio) more loudly and then when mic action/voice stops for lets say 3-5 seconds the volume of original audio stream fades up/goes up to normal volume... is this too much, i can play with sox or similar ?