Recherche avancée

Sur d’autres sites (10910)

• ffmpeg generate overlay complex filter for audio with background image

  8 janvier 2016, par user1793606

  I am experimenting with ffmpeg and would like to generate overlay complex filter for audio with background image. The code normally works, except for when adding -filter_complex "[0:a]showwaves=s=1280x720:mode=line,format=yuv420p[v]" -map "[v]" -map 0:a
  it crashes. I found the example code at https://trac.ffmpeg.org/wiki/Waveform Any help is appreciated.

  Command '['c:/ffmpeg/bin\\ffmpeg.exe', '-y', '-loop', '1', '-r', '1', '-i', 'temp\\bg.png', '-i', 'test.mp3', '-filter_complex', '[0:a]showwaves=s=1280x720:mode=line,format=yuv420p[v]', '-map', '[v]', '-map', '0:a', '-c:v', 'libx264', '-preset', 'ultrafast', '-tune', 'stillimage', '-crf', '15', '-pix_fmt', 'yuv420p', '-strict', 'experimental', '-c:a', 'aac', '-b:a', '256k', '-shortest', '-threads', '0', 'done/test.mp4']'

  EDIT 1 :

  I tested this with the new verson : ffmpeg -y -i input.mp3 -i background.png -filter_complex "[0:a]showwaves=s=1280x720:mode=line,format=yuv420p[v]" -map "[v]" -map 0:a -c:v libx264 -c:a copy output.mp4

  It generates an output file, but only the waveform, no background included. My end goal is to generate the waveform over the background.

• using pocketsphinx_continuous with a .wav file

  3 avril 2013, par user2242131

  I am attempting to write an application that will allow a user to speak a small set of commands from a remote system and have them executed on my server. Using pocketsphinx to parse the spoken text. When run locally with the microphone, pocketsphinx_continuous works perfectly no matter how I slur the words. But when importing the audio file and using ffmpeg to downsample the audio to a single channel, 16 bit PCM file, it will parse the first word without difficulty. Then it will skip everything else and treat it as . I am confident that the problem is in the file format and not in the pocketsphinx configuration.

  Using command line
  ffmpeg -y -i Sound\AddSheet.wav -ac 1 -f s16le -acodec pcm_s16le -ar 16k AddTmp.wav
  in a batch file.

  The bottom of the output I get is :

  INFO: fsg_search.c(1407): Start node ADD.0:5:47
  
  INFO: fsg_search.c(1407): Start node <sil>.0:2:49
  
  INFO: fsg_search.c(1446): End node <sil>.126:128:305 (-486)
  
  INFO: fsg_search.c(1662): lattice start node <s>.0 end node <sil>.126
  
  INFO: ps_lattice.c(1352): Normalizer P(O) = alpha(<sil>:126:305) = -175371
  
  INFO: ps_lattice.c(1390): Joint P(O,S) = -176076 P(S|O) = -705
  
  000000000: ADD USER
  
  </sil></sil></s></sil></sil>

  Which is not the audio in the file. The words spoken in the file are "ADD SPREADSHEET", which works perfectly from the same microphone without the intervening .wav file.

  I have tried increasing the audio volume and decreasing the background noise using sox :

  sox -v 3.0 Sound\%1 Sound\%1-loud.wav ffmpeg -i Sound\%1-loud.wav -vn -ss 00:00:00 -t 00:00:01 -y Sound\%1-noiseaud.wav 
  
  sox Sound\%1-noiseaud.wav -n noiseprof Sound\%1-noise.prof 
  
  sox Sound\%1 Sound\%1-clean.wav noisered sound\noise.prof 0.21 
  
  ffmpeg -y -i Sound\%1-clean.wav -ac 1 -f s16le -acodec pcm_s16le -ar 16k AddTmp.wav

  with no noticeable effect on the final results.

  If you look at the output you will notice that fsg_search.c has found ADD as the start node, then silence for the remainder. Please help on this.

• cbs_mpeg2 : Improve checks for invalid values

  22 mai 2019, par Andreas Rheinhardt

  cbs_mpeg2 : Improve checks for invalid values
  MPEG-2 contains several elements that mustn't be zero according to the
  
  specifications : horizontal/vertical_size_value, aspect_ratio_information,
  
  frame_rate_code, the quantiser matrices, the colour_description
  
  elements, picture_coding_type, the f_code[r][s] values and
  
  quantiser_scale_code. It is now checked that the invalid values don't
  
  occur.
  The colour_description elements are treated specially in this regard :
  
  Given that there are files in the wild which use illegal values for the
  
  colour_description elements (some of them created by mpeg2_metadata),
  
  they will be corrected to the value meaning "unknown" (namely 2) during
  
  reading. This has been done in such a way that trace_headers will
  
  nevertheless report the original value, together with a message about
  
  the fixup.
  Furthermore, the trace_headers output of user_data has been beautified.
  Signed-off-by : Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
  

  • [DH] libavcodec/cbs_mpeg2.c
  • [DH] libavcodec/cbs_mpeg2_syntax_template.c