Recherche avancée

Sur d’autres sites (10664)

• How to use FFMPEG for a cutaway with audio crossfade ?

  14 mai 2016, par jreikes

  I’m trying to do a "cutaway" (change camera angles) using FFMPEG. I have HLS video from a left camera and a right camera. The start times are a little out of sync, so the left video actually started recording 4.21 seconds after the right video started recording. I’m trying to show the right video from the start to the 5 second mark and then cut over to the left video from its 0.79 second mark until the 5.79 second mark (which would have been the 5-10 second range on the right camera).

  I realize these cuts probably fall on p-frames — that’s part of the test here. I want to see if FFMPEG will properly insert the necessary i-frame to make a smooth cut.

  Finally, to get a smooth final product, I’m trying to do a 1-second audio crossfade during the transition (there’s no need for any video fades).

  I’m new at FFMPEG. I tried the following, but it didn’t work. Anyone know what the correct command would have been ?

  ffmpeg -i right.m3u8 -ss 0 -t 5 -i left.m3u8 -ss 0.79 -t 5 -async 1 -filter_complex acrossfade=d=1:c1=exp:c2=exp mixed.m3u8

• Why are some delay load DLLs immediately unloaded with FFMPEG ?

  21 avril 2020, par alvion

  I've built ffmpeg as DLLs and linked them to my application, which runs fine when the DLLs are in the executable folder. Because I would like to load different builds of ffmpeg for different project configurations, I need to put them in different folders than the .exe (ie "/bin/ffmpeg/config1", "/bin/ffmpeg/config2", etc). The only way I know how to do this without modifying the environment PATH is to mark the DLLs as "delay load" and then specify the full path via LoadLibrary calls at startup.

  



  This works fine for other DLLs, but for FFMPEG I'm having trouble getting it to work properly.

  



  I've specified the DLLs to DelayLoad in my vcxproj (actually in a linked .props file) :

  



    <delayloaddlls>&#xA;    avcodec-58.dll;avutil-56.dll;avformat-58.dll;swscale-5.dll;swresample-3.dll;&#xA;  </delayloaddlls>&#xA;

  



  And at startup, I'm loading them via LoadLibrary :

  



  LoadLibraryA("ffmpeg\\config1\\avcodec-58.dll");
LoadLibraryA("ffmpeg\\config1\\avutil-56.dll");
LoadLibraryA("ffmpeg\\config1\\avformat-58.dll");
LoadLibraryA("ffmpeg\\config1\\swscale-5.dll");
LoadLibraryA("ffmpeg\\config1\\swresample-3.dll");


  



  The strange thing is, when I make the LoadLibrary calls, it seems to be loading them and then unloading some of them immediately :

  



  'ffmpegtest.exe' (Win32): Loaded 'C:\ffmpegtest\bin\ffmpeg\medium\avcodec-58.dll'. Symbols loaded.
'ffmpegtest.exe' (Win32): Unloaded 'C:\ffmpegtest\bin\ffmpeg\medium\avcodec-58.dll'
'ffmpegtest.exe' (Win32): Loaded 'C:\ffmpegtest\bin\ffmpeg\medium\avutil-56.dll'. Symbols loaded.
'ffmpegtest.exe' (Win32): Loaded 'C:\ffmpegtest\bin\ffmpeg\medium\avformat-58.dll'. Symbols loaded.
'ffmpegtest.exe' (Win32): Unloaded 'C:\ffmpegtest\bin\ffmpeg\medium\avformat-58.dll'
'ffmpegtest.exe' (Win32): Loaded 'C:\ffmpegtest\bin\ffmpeg\medium\swscale-5.dll'. Symbols loaded.
'ffmpegtest.exe' (Win32): Loaded 'C:\ffmpegtest\bin\ffmpeg\medium\swresample-3.dll'. Symbols loaded.


  



  So, the problem DEFINITELY seems to be with the fact that it is choosing to unload some of the dlls. If I don't include any calls at all to LoadLibrary, then I get a crash in the very first line of code calling into these libraries, a function in avutil-56.dll. If I include the calls to LoadLibrary, then the first couple function calls succeed, and then we crash in the first call to an unloaded library, specifically "av_codec_iterate()" in avcodec-58.dll. If I copy the unloaded DLLs to the executable folder, then everything runs fine.

  



  Why are some of the DLLs being unloaded ? And how do I prevent it ?

  


• ffmpeg get the audio stream from mp4 and send it to speech recognition

  12 juillet 2013, par user1896859

  I have few .mp4 video files in which at the start of each video file there is a word, I want to load these files get the audio check what is the spoken word and rename the file accordingly.

  Currently what i am doing is, converting all the mp4 files to wav and then sending the to speech recognition and then doing the renaming stuff.

  Is there a way to cut short the "converting to wav" part out and directly send the mp4 audio stream to speech recognition ??

  Thanks,