[FFmpeg-user] Keeping A/V sync during telecine

[Carl Zwanzig]

Hello,

In order to support a legacy analog NTSC system, I need to convert a pile of 
programs from 24p to 60i with a 3:2 pulldown. I've been successful with the 
video, however sometimes the audio comes out early. In some test cases, it's 
noticeably off even a couple of minutes into program; can't really tell if 
it drifts further since it's not in sync to start. (And, of course, some 
programs convert just fine.)


Example:
cpz at video-proc-0:/mnt/videofiles % ffmpeg -r ntsc-film -i INPUT.vob -vf 
telecine -dc 9 -target ntsc-dvd -flags +ilme+ildct -alternate_scan 1 -top 0 
OUTPUT.mpg
ffmpeg version 3.3.5 Copyright (c) 2000-2017 the FFmpeg developers
   built with FreeBSD clang version 4.0.0 (tags/RELEASE_400/final 297347) 
(based on LLVM 4.0.0)
   configuration: --prefix=/usr/local --mandir=/usr/local/man 
--datadir=/usr/local/share/ffmpeg 
--pkgconfigdir=/usr/local/libdata/pkgconfig --enable-shared --enable-pic 
--enable-gpl --enable-postproc --enable-avfilter --enable-avresample 
--enable-pthreads --cc=cc --disable-indev=alsa --disable-outdev=alsa 
--disable-libopencore-amrnb --disable-libopencore-amrwb --disable-libass 
--disable-libbs2b --disable-libcaca --disable-libcdio --disable-libcelt 
--disable-chromaprint --disable-libdc1394 --disable-debug --enable-htmlpages 
--disable-libfdk-aac --enable-ffserver --disable-libflite 
--enable-fontconfig --enable-libfreetype --enable-frei0r 
--disable-libfribidi --disable-libgme --disable-libgsm --enable-iconv 
--disable-libilbc --disable-indev=jack --disable-libkvazaar --disable-ladspa 
--disable-libmp3lame --disable-libbluray --enable-mmx --disable-libmodplug 
--disable-netcdf --disable-openal --disable-indev=openal --disable-opencl 
--enable-libopencv --disable-opengl --disable-libopenh264 
--disable-libopenjpeg --enable-optimizations --disable-libopus 
--disable-libpulse --disable-indev=pulse --disable-outdev=pulse 
--enable-runtime-cpudetect --disable-librubberband --enable-libschroedinger 
--disable-ffplay --disable-outdev=sdl --disable-sdl2 --disable-libsmbclient 
--disable-libsnappy --disable-indev=sndio --disable-outdev=sndio 
--disable-libsoxr --disable-libspeex --enable-sse --disable-libssh 
--disable-libtesseract --enable-libtheora --disable-libtwolame 
--enable-libv4l2 --enable-vaapi --enable-vdpau --disable-libvidstab 
--enable-libvorbis --disable-libvo-amrwbenc --enable-libvpx 
--disable-libwavpack --disable-libwebp --enable-libx264 --enable-libx265 
--disable-libxcb --enable-libxvid --disable-outdev=xv --disable-libzimg 
--disable-libzmq --disable-libzvbi --disable-gcrypt --enable-gmp 
--disable-librtmp --enable-gnutls --disable-openssl --enable-version3 
--disable-nonfree
   libavutil      55. 58.100 / 55. 58.100
   libavcodec     57. 89.100 / 57. 89.100
   libavformat    57. 71.100 / 57. 71.100
   libavdevice    57.  6.100 / 57.  6.100
   libavfilter     6. 82.100 /  6. 82.100
   libavresample   3.  5.  0 /  3.  5.  0
   libswscale      4.  6.100 /  4.  6.100
   libswresample   2.  7.100 /  2.  7.100
   libpostproc    54.  5.100 / 54.  5.100
Input #0, mpeg, from 'INPUT.vob':
   Duration: 01:55:53.25, start: 0.084000, bitrate: 5061 kb/s
     Stream #0:0[0x1bf]: Data: dvd_nav_packet
     Stream #0:1[0x1e0]: Video: mpeg2video (Main), yuv420p(tv, bt470bg, top 
first), 720x480 [SAR 8:9 DAR 4:3], 29.97 fps, 59.94 tbr, 90k tbn, 59.94 tbc
     Stream #0:2[0x80]: Audio: ac3, 48000 Hz, stereo, fltp, 256 kb/s
Stream mapping:
   Stream #0:1 -> #0:0 (mpeg2video (native) -> mpeg2video (native))
   Stream #0:2 -> #0:1 (ac3 (native) -> ac3 (native))
Press [q] to stop, [?] for help
[Parsed_telecine_0 @ 0x80c23e920] Telecine pattern 23 yields up to 2 frames 
per frame, pts advance factor: 4/5
Output #0, dvd, to 'OUTPUT.mpg':
   Metadata:
     encoder         : Lavf57.71.100
     Stream #0:0: Video: mpeg2video (Main), yuv420p(progressive), 720x480 
[SAR 8:9 DAR 4:3], q=2-31, 6000 kb/s, 29.97 fps, 90k tbn, 29.97 tbc
     Metadata:
       encoder         : Lavc57.89.100 mpeg2video
     Side data:
       cpb: bitrate max/min/avg: 9000000/0/6000000 buffer size: 1835008 
vbv_delay: -1
     Stream #0:1: Audio: ac3, 48000 Hz, stereo, fltp, 448 kb/s
     Metadata:
       encoder         : Lavc57.89.100 ac3
frame=208813 fps=437 q=2.0 Lsize= 4942192kB time=01:56:07.32 
bitrate=5810.9kbits/s speed=14.6x
video:4470286kB audio:380256kB subtitle:0kB other streams:0kB global 
headers:0kB muxing overhead: 1.889482%
cpz at video-proc-0:/mnt/videofiles/ %

Adding the filter "-af aresample=async=1000" did not seem to change the 
behavior, nor did "-async 1".

Many of the originals I've been given have some sort of silent title/intro, 
so I wonder whether actual audio packets don't start immediately and that 
something isn't matching or adjusting timestamps. (Sync issues are seen with 
multiple players and original is fine with vlc.)

I need some ideas :).

Manually adjusting an audio delay isn't a great option as I have 50-60 
individual items to convert, and I'd rather set a batch running and leave 
it. Also, I'm not overly concerned with artifacts, either visual or aural, 
or against encoding into h264 as part of the process. Whatever will work.

Many thanks,

z!