Context Navigation

#1755 closed defect (fixed)

Incorrect timestamps for aac in avi

Reported by:	Heesuk Jung	Owned by:
Priority:	normal	Component:	avformat
Version:	git-master	Keywords:	avi aac pts
Cc:	dwkang@google.com, myungchul.kim@lge.com, suhwang.kim@lge.com, woojun.jung@lge.com, donggul.kim@lge.com, hojae.lee@lge.com, kubong.min@lge.com	Blocked By:
Blocking:		Reproduced by developer:	yes
Analyzed by developer:	no

Description

Summary of the bug:

Audio become silent before playback reach end of file.
Video end at about 31 sec but audio stopped about 12 sec.

I reviewed audio and video PTS information using ffprobe tool.
Please refer the PTS plotting graph as below.
(Notice! Recommend download of PTS plotting file because of Google docs excel inter-operability)

PTS plotting link : https://docs.google.com/open?id=0B6r7ZfWFIypCbnRtVlFTVVA0cEU
Problematic file link : https://docs.google.com/open?id=0B6r7ZfWFIypCeXB3b2J5S1NDMzQ

My Analysis of PTS

Video PTS is OK.
Audio PTS is NG.
- Whole timeline of this media file is about 31 sec but audio pts has about 12 sec timeline.
- Also values of Audio PTS are grouped as special pattern.
- Simplified Audio PTS pattern : 0, 0, 0, 1, 1, 2, 2, ……. , N/3
- Simplified Video PTS pattern : 0, 1, 2, …. , N

-> The shape of audio PTS make 2 problem.

1) Audio stop early. Because audio PTS by ffmpeg have 12 sec but real audio has 30 sec.
2) Seek does not work after 12 sec.

My co-worker checked audio PTS information of problematic based on avidemux and Audio PTS is just OK.
And so I request that FFmpeg team inspect this problem.

How to reproduce:
% ffplay 016_avi_mpeg4_sp[720@24fps]_aac[2ch_128kbps_44.1khz].avi[MV]_origin.avi

Attachments (5)

avidec.c (54.0 KB ) - added by Heesuk Jung 12 years ago.: suggest using sample size to improve current issue
frameInfo_withoutPatch.txt (569.2 KB ) - added by Heesuk Jung 12 years ago.: Original Frame information based on ffprobe
frameInfo_withPatch.txt (571.5 KB ) - added by Heesuk Jung 12 years ago.: Patch applied frame information based on ffprobe
avidec.patch (3.3 KB ) - added by Heesuk Jung 12 years ago.: git format-patch file
0001-PATCH-Use-sample-size-in-case-incorrect-timestamps-f.patch (1.4 KB ) - added by Heesuk Jung 12 years ago.: update patch considering additional problem case

Download all attachments as: .zip

Change History (16)

comment:1 by Carl Eugen Hoyos, 12 years ago

Keywords:	Grouped Audio PTS removed

Is this also reproducible with current git head?

To make this a valid ticket, please add command line together with complete, uncut console output.

comment:2 by Carl Eugen Hoyos, 12 years ago

Analyzed by developer:	unset
Reproduced by developer:	unset

comment:3 by Heesuk Jung, 12 years ago

I update complete console output as below.

shkim81@GuHaRa:~/ffmpeg-0.11.1$ ./ffplay ~/AVI/016_avi_mpeg4_sp\[720@24fps\]_aac\[2ch_128kbps_44.1khz\].avi\[MV\]_origin.avi 
ffplay version 0.11.1 Copyright (c) 2003-2012 the FFmpeg developers
  built on Jun 27 2012 21:31:24 with gcc 4.4.3
  configuration: --prefix=/home/2.Study/1.ffmepg/ffmpeg --disable-yasm --enable-shared
  libavutil      51. 54.100 / 51. 54.100
  libavcodec     54. 23.100 / 54. 23.100
  libavformat    54.  6.100 / 54.  6.100
  libavdevice    54.  0.100 / 54.  0.100
  libavfilter     2. 77.100 /  2. 77.100
  libswscale      2.  1.100 /  2.  1.100
  libswresample   0. 15.100 /  0. 15.100
[avi @ 0x23b7ce0] non-interleaved AVI
Input #0, avi, from '/home/shkim81/MediaDB_GCD/1_SIC_USB/CompatibilityFullPackage/AVI/016_avi_mpeg4_sp[720@24fps]_aac[2ch_128kbps_44.1khz].avi[MV]_origin.avi':
  Metadata:
    encoder         : PotEncoder
  Duration: 00:00:31.00, start: 0.000000, bitrate: 6907 kb/s
    Stream #0:0: Video: mpeg4 (Simple Profile) (XVID / 0x44495658), yuv420p, 1280x720 [SAR 1:1 DAR 16:9], 24 tbr, 24 tbn, 24 tbc
    Stream #0:1: Audio: aac ([255][0][0][0] / 0x00FF), 44100 Hz, stereo, s16, 128 kb/s
  36.69 A-V: -0.021 fd=   5 aq=    0KB vq=    0KB sq=    0B f=0/0   f=0/0

Last edited 12 years ago by Carl Eugen Hoyos (previous) (diff)

comment:4 by Carl Eugen Hoyos, 12 years ago

Don't you agree that the problem can be much better seen in ffplay's output if you stop playback after ~11 seconds?

Please test current git head.

comment:5 by Heesuk Jung, 12 years ago

Hi cehoyos,

I don't think so because a/v playback of other player such as Avidemux works well until end of stream.
Also avidemux extracts the complete audio pts information of problematic files.
I recommend that you play the probemlatic file using avidemux player.

link : http://avidemux.berlios.de/download.html

Test result of ffmpeg latest mater branch is same as ffmpeg 0.11.1.

hsjung@hsjung-laptop:~/utils/ffmpeg$ ./ffplay /media/LGEXT/Sanity_Test/1_SIC_USB/CompatibilityFullPackage/AVI/016_avi_mpeg4_sp\[720@24fps\]_aac\[2ch_128kbps_44.1khz\].avi\[MV\]_origin.avi 
ffplay version git-2012-09-20-16e52c8 Copyright (c) 2003-2012 the FFmpeg developers
  built on Sep 20 2012 18:00:20 with gcc 4.4.3 (Ubuntu 4.4.3-4ubuntu5.1)
  configuration: --enable-gpl --enable-libfaac --enable-libmp3lame --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libtheora --enable-libvorbis --enable-libvpx --enable-libx264 --enable-nonfree --enable-version3 --enable-x11grab
  libavutil      51. 73.101 / 51. 73.101
  libavcodec     54. 56.100 / 54. 56.100
  libavformat    54. 27.101 / 54. 27.101
  libavdevice    54.  2.100 / 54.  2.100
  libavfilter     3. 16.104 /  3. 16.104
  libswscale      2.  1.101 /  2.  1.101
  libswresample   0. 15.100 /  0. 15.100
  libpostproc    52.  0.100 / 52.  0.100
[avi @ 0x26eb3e0] non-interleaved AVI
Input #0, avi, from '/media/LGEXT/Sanity_Test/1_SIC_USB/CompatibilityFullPackage/AVI/016_avi_mpeg4_sp[720@24fps]_aac[2ch_128kbps_44.1khz].avi[MV]_origin.avi':
  Metadata:
    encoder         : PotEncoder
  Duration: 00:00:31.00, start: 0.000000, bitrate: 6907 kb/s
    Stream #0:0: Video: mpeg4 (Simple Profile) (XVID / 0x44495658), yuv420p, 1280x720 [SAR 1:1 DAR 16:9], 24 tbr, 24 tbn, 24 tbc
    Stream #0:1: Audio: aac ([255][0][0][0] / 0x00FF), 44100 Hz, stereo, s16, 128 kb/s
Frame changed from size:0x0 to size:1280x720 vq=  205KB sq=    0B f=0/0   
  31.56 A-V: -0.038 fd=   0 aq=    0KB vq=    0KB sq=    0B f=0/0   f=0/0   
hsjung@hsjung-laptop:~/utils/ffmpeg$

Thanks

Last edited 12 years ago by Carl Eugen Hoyos (previous) (diff)

comment:6 by Carl Eugen Hoyos, 12 years ago

Keywords:	avi aac pts added
Reproduced by developer:	set
Status:	new → open
Summary:	Audio becomes silent on some avi files. (This problem is reproduced during 1x play and seek) → Incorrect timestamps for aac in avi
Version:	0.11.1 → git-master

Audio timestamps for the sample file (uploaded to samples/ffmpeg-bugs/trac/ticket1755) are wrong with lavf, ffplay tries to correct it by playing video too slow, leading to A/V-desync, same wrong timestamps are shown with mplayer -demuxer lavf, the file plays fine with mplayer -demuxer avi

$ time ffplay 016_avi_mpeg4_sp\[720\@24fps\]_aac\[2ch_128kbps_44.1khz\].avi\[MV\]_origin.avi
ffplay version N-44632-ga16c512 Copyright (c) 2003-2012 the FFmpeg developers
  built on Sep 20 2012 16:11:00 with gcc 4.5.3 (GCC)
  configuration: --cc=/usr/local/gcc-4.5.3/bin/gcc --enable-gpl
  libavutil      51. 73.101 / 51. 73.101
  libavcodec     54. 56.100 / 54. 56.100
  libavformat    54. 27.101 / 54. 27.101
  libavdevice    54.  2.101 / 54.  2.101
  libavfilter     3. 16.105 /  3. 16.105
  libswscale      2.  1.101 /  2.  1.101
  libswresample   0. 15.100 /  0. 15.100
  libpostproc    52.  0.100 / 52.  0.100
[avi @ 0x151eb20] non-interleaved AVI
Input #0, avi, from '016_avi_mpeg4_sp[720@24fps]_aac[2ch_128kbps_44.1khz].avi[MV]_origin.avi':
  Metadata:
    encoder         : PotEncoder
  Duration: 00:00:31.00, start: 0.000000, bitrate: 6907 kb/s
    Stream #0:0: Video: mpeg4 (Simple Profile) (XVID / 0x44495658), yuv420p, 1280x720 [SAR 1:1 DAR 16:9], 24 tbr, 24 tbn, 24 tbc
    Stream #0:1: Audio: aac ([255][0][0][0] / 0x00FF), 44100 Hz, stereo, s16, 128 kb/s
Frame changed from size:0x0 to size:1280x720
   9.98 A-V: -2.689 fd=   0 aq=  118KB vq=  233KB sq=    0B f=0/0   f=0/0

real    0m25.429s
user    0m2.105s
sys     0m0.218s

ffmpeg corrects the timestamps, resulting video seems fine afaict:

$ ffmpeg -i 016_avi_mpeg4_sp\[720\@24fps\]_aac\[2ch_128kbps_44.1khz\].avi\[MV\]_origin.avi -qscale 2 out.avi
ffmpeg version N-44632-ga16c512 Copyright (c) 2000-2012 the FFmpeg developers
  built on Sep 20 2012 16:11:00 with gcc 4.5.3 (GCC)
  configuration: --cc=/usr/local/gcc-4.5.3/bin/gcc --enable-gpl
  libavutil      51. 73.101 / 51. 73.101
  libavcodec     54. 56.100 / 54. 56.100
  libavformat    54. 27.101 / 54. 27.101
  libavdevice    54.  2.101 / 54.  2.101
  libavfilter     3. 16.105 /  3. 16.105
  libswscale      2.  1.101 /  2.  1.101
  libswresample   0. 15.100 /  0. 15.100
  libpostproc    52.  0.100 / 52.  0.100
[avi @ 0x14f0240] non-interleaved AVI
Input #0, avi, from '016_avi_mpeg4_sp[720@24fps]_aac[2ch_128kbps_44.1khz].avi[MV]_origin.avi':
  Metadata:
    encoder         : PotEncoder
  Duration: 00:00:31.00, start: 0.000000, bitrate: 6907 kb/s
    Stream #0:0: Video: mpeg4 (Simple Profile) (XVID / 0x44495658), yuv420p, 1280x720 [SAR 1:1 DAR 16:9], 24 tbr, 24 tbn, 24 tbc
    Stream #0:1: Audio: aac ([255][0][0][0] / 0x00FF), 44100 Hz, stereo, s16, 128 kb/s
Please use -q:a or -q:v, -qscale is ambiguous
Output #0, avi, to 'out.avi':
  Metadata:
    ISFT            : Lavf54.27.101
    Stream #0:0: Video: mpeg4 (FMP4 / 0x34504D46), yuv420p, 1280x720 [SAR 1:1 DAR 16:9], q=2-31, 200 kb/s, 24 tbn, 24 tbc
    Stream #0:1: Audio: ac3 ([0] [0][0] / 0x2000), 44100 Hz, stereo, flt, 192 kb/s
Stream mapping:
  Stream #0:0 -> #0:0 (mpeg4 -> mpeg4)
  Stream #0:1 -> #0:1 (aac -> ac3)
Press [q] to stop, [?] for help
[avi @ 0x150b000] st:0 PTS: 3 DTS: 3 < 6 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 3 DTS: 3 < 7 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 4 DTS: 4 < 8 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 5 DTS: 5 < 9 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 6 DTS: 6 < 10 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 5 DTS: 5 < 11 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 6 DTS: 6 < 12 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 7 DTS: 7 < 13 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 7 DTS: 7 < 14 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 7 DTS: 7 < 15 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 8 DTS: 8 < 16 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 8 DTS: 8 < 17 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 9 DTS: 9 < 18 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 9 DTS: 9 < 19 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 10 DTS: 10 < 20 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 9 DTS: 9 < 21 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 10 DTS: 10 < 22 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 11 DTS: 11 < 23 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 11 DTS: 11 < 24 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 11 DTS: 11 < 25 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 12 DTS: 12 < 26 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 12 DTS: 12 < 27 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 13 DTS: 13 < 28 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 13 DTS: 13 < 29 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 14 DTS: 14 < 30 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 14 DTS: 14 < 31 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 14 DTS: 14 < 32 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 15 DTS: 15 < 33 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 15 DTS: 15 < 34 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 15 DTS: 15 < 35 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 16 DTS: 16 < 36 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 16 DTS: 16 < 37 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 17 DTS: 17 < 38 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 17 DTS: 17 < 39 invalid, clipping

...

[avi @ 0x150b000] st:0 PTS: 341 DTS: 341 < 861 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 341 DTS: 341 < 862 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 341 DTS: 341 < 863 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 342 DTS: 342 < 864 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 342 DTS: 342 < 865 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 343 DTS: 343 < 866 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 343 DTS: 343 < 867 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 344 DTS: 344 < 868 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 343 DTS: 343 < 869 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 344 DTS: 344 < 870 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 344 DTS: 344 < 871 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 345 DTS: 345 < 872 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 345 DTS: 345 < 873 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 346 DTS: 346 < 874 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 346 DTS: 346 < 875 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 346 DTS: 346 < 876 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 347 DTS: 347 < 877 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 347 DTS: 347 < 878 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 347 DTS: 347 < 879 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 348 DTS: 348 < 880 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 349 DTS: 349 < 881 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 349 DTS: 349 < 882 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 349 DTS: 349 < 883 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 350 DTS: 350 < 884 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 350 DTS: 350 < 885 invalid, clipping
[avi @ 0x150b000] st:0 PTS: 350 DTS: 350 < 886 invalid, clipping
frame=  744 fps= 94 q=2.0 Lsize=   36906kB time=00:00:31.00 bitrate=9752.7kbits/s
video:36133kB audio:725kB subtitle:0 global headers:0kB muxing overhead 0.131191%

comment:7 by Heesuk Jung, 12 years ago

I have already understood that audio timestamp extracted lavf was wrong as previous linked pts plotting.
I wonder if ffmpeg has some idea to improve this problem or not.

Actually, I found that 1x playback is ok using decoder free run mode.
(Decoder decide pts information without receiving parser pts.)
But I knew that it is so hard to improve seek problem.
FFmpeg parser get wrong frames because of wrong pts information.
At least, I hope to find solution for seek problem.

Thanks.

by Heesuk Jung, 12 years ago

Attachment:	avidec.c added

suggest using sample size to improve current issue

by Heesuk Jung, 12 years ago

Attachment:	frameInfo_withoutPatch.txt added

Original Frame information based on ffprobe

by Heesuk Jung, 12 years ago

Attachment:	frameInfo_withPatch.txt added

Patch applied frame information based on ffprobe

comment:8 by Heesuk Jung, 12 years ago

Hi cehoyos,

I fixed this issue.
Please refer my summary as below.

Use sample size in case incorrect timestamps for aac in AVI (Ticket #1755)

In some case for aac in AVI, avidec extracts wrong PTS value.
(www.ffmpeg.org/trac/ffmpeg/ticket/1755)
In getDuration function, I suggest using sample size instead of len.

I attached 4 files.

avidec.c
- In latest master, I merged my patch into avidec.c
avidec.patch
- patch file
frameInfo_withoutPatch.txt
- There are frame informations without patch. As you know, PTS value is wrong.
- 1X playback is NG and seek is NG too.
frameInfo_withPatch.txt
- There are frame informations without patch. As you know, PTS value is ok.
- 1X playback is ok and seek is ok too.

Also I have done regression test in our media pool.

Test result of 11 problematic files is ok after patch apply.
(11 files have fixed sample size in AVI Stream Header)

Regression test reseult of 4 non-problematic files is ok after patch apply.
(4 files have variable sample size in AVI Stream Header)

I have 1 question to commit change.
I got ffmpeg master branch but master git is read-only and so
I hope to know how I can submit change into master branch?
(I want to know change commit process or needing some admission.)

Thanks

comment:9 by Carl Eugen Hoyos, 12 years ago

Please send your patch to ffmpeg-devel, patches generally receive more attention there.

by Heesuk Jung, 12 years ago

Attachment:	avidec.patch added

git format-patch file

comment:10 by Carl Eugen Hoyos, 12 years ago

Resolution:	→ fixed
Status:	open → closed

Fixed by Michael, please report if some files still fail for you!

by Heesuk Jung, 12 years ago

Attachment:	0001-PATCH-Use-sample-size-in-case-incorrect-timestamps-f.patch added

update patch considering additional problem case

comment:11 by Heesuk Jung, 12 years ago

I found more problem case and so add condition based on Michael patch.

Thanks!

Note: See TracTickets for help on using tickets.

Download in other formats: