ffmpeg video conversion issue

[hss2293]

I have converted a video using ffmpeg.. The details of the original files was Duration: 01:31:40.81, start: 0.000000
but when i converted the video the info of converted file was Duration: 01:31:40.85, start: 0.046440
After converting the video the start: 0.000000 changed to start: 0.046440
What does it mean ? Why it got changed ? How to make both same even after converting ?

[hss2293]

Please reply if someone has knowledge about it..

[Budman1]

I've noticed this a lot and wrote a program to show a visual indication of audio to video sync because of that but it can be caused by a LOT of things like:
1. Audio delay with to respect to video as shown by MediaInfo
2. Audio and Video being resynced like:

VVVVVVVVVVVVVV
AAAAAAAAAAAAAA (shifted .046 -> 01:31:40.81 + 00:00:00.046 = 01:31:40.856 Total V-A)

3. MediaInfo reading Metadata not the actual settings (Which it does occasionally if Meta is present)

Without actually being sure where you came from and what format, codec, information utility you used, etc. it is hard to tell.
Since the result resembles item 2 above, I'm betting on that one. Without a lot of testing, it will be equally hard to tell if that's what SHOULD have happened.

Is the Video and Audio in sync? It will be hard to tell with only .046 shift.

[poisondeathray]

Audio is padded by many encoders e.g. AAC by ~ 40ms. The value varies by sample rate and encoder. The delay in the video should keeps everything in sync (40ms is about the threshold that most people can discern anyway)

[Selur]

Depending of what the input is all about, there might also be some not decodeable frames which get ignored during reencoding and might lead to a delay.
Without infos about the source and the used ffmpeg call there are to many variables that might be the cause of the delay,...

[hss2293]

Yes the audio and video are in sync.. But i want to solve this issue.. I have seen many videos on other sites and tested them they all have start: 0.000000..
Is there any way to fix it ?

[DB83]

Yes the audio and video are in sync.. But i want to solve this issue.. I have seen many videos on other sites and tested them they all have start: 0.000000..
Is there any way to fix it ?

What is there to fix ? The audio is in sync so that really is all that should concern you.

But if it really bothers you, you should at the very least provide the information that those who have been generous enough to reply have requested.

[Budman1]

Can you post the mediainfo from it so we at least know what file type it is? There are ways to get ffmpeg to 'fix' that but it depends on the codec and container, the length of each audio and video, audio delay, etc. Thanks

[hss2293]

I am using the following command

ffmpeg -i filename -vcodec mpeg4 -b:v 170k -s 320x240 -acodec libfdk_aac  -b:a 32k -ar 44100 -ac 1 output.3gp

Its giving me delay of start: 0.046440 .. This delay is everytime i convert video of any duration..

[Budman1]

Cqn you post the MediaInfo of the first original video. This is important since it may have original delay, uneven audio/video lengths, etc. If as I suspect, the audio/video needs to be offset, you might try taking the resulting video and run 'ffmpeg -I video.mp4 -ss 0.046 -vcodec copy -acodec copy video2.mp4. This will essentially 'trim' any projecting extra 0.046 off the front and should eliminate the need for any delay. Its worth a try and has fixed many of mine but since I do not have the original or mediainfo, we can only guess. The only way to truly tell ia to separate the audio and video to see if they are ok (equal length).

[hss2293]

Here is mediainfo of original video.

Format                                   : MPEG-4
Format profile                           : Base Media / Version 2
Codec ID                                 : mp42
File size                                : 51.4 MiB
Duration                                 : 4mn 55s
Overall bit rate mode                    : Variable
Overall bit rate                         : 1 459 Kbps
Encoded date                             : UTC 2013-11-26 22:58:29
Tagged date                              : UTC 2013-11-26 22:58:29
gsst                                     : 0
gstd                                     : 295333
gssd                                     : BADC207E2HH1386080592668708
gshh                                     : r6---sn-p5qlsne7.googlevideo.com

Video
ID                                       : 1
Format                                   : AVC
Format/Info                              : Advanced Video Codec
Format profile                           : High@L3.1
Format settings, CABAC                   : Yes
Format settings, ReFrames                : 1 frame
Codec ID                                 : avc1
Codec ID/Info                            : Advanced Video Coding
Duration                                 : 4mn 55s
Bit rate                                 : 1 265 Kbps
Maximum bit rate                         : 4 129 Kbps
Width                                    : 1 280 pixels
Height                                   : 720 pixels
Display aspect ratio                     : 16:9
Frame rate mode                          : Constant
Frame rate                               : 25.000 fps
Color space                              : YUV
Chroma subsampling                       : 4:2:0
Bit depth                                : 8 bits
Scan type                                : Progressive
Bits/(Pixel*Frame)                       : 0.055
Stream size                              : 44.5 MiB (87%)
Tagged date                              : UTC 2013-11-26 22:58:33

Audio
ID                                       : 2
Format                                   : AAC
Format/Info                              : Advanced Audio Codec
Format profile                           : LC
Codec ID                                 : 40
Duration                                 : 4mn 55s
Bit rate mode                            : Variable
Bit rate                                 : 192 Kbps
Maximum bit rate                         : 203 Kbps
Channel(s)                               : 2 channels
Channel positions                        : Front: L R
Sampling rate                            : 44.1 KHz
Compression mode                         : Lossy
Stream size                              : 6.76 MiB (13%)
Title                                    : IsoMedia File Produced by Google, 5-11-2011
Encoded date                             : UTC 2013-11-26 22:58:31
Tagged date                              : UTC 2013-11-26 22:58:33

[Budman1]

That video has been downloaded, at some time, from YouTube and is variable bit rate for audio and video. I did a test and downloaded a similar video from YouTube and got the same results as you. My audio was 10ms off instead of 46. I tried the same ffmpeg script as yours except used libvo_aacenc for audio and this time it simply made the last audio frame 10ms in length to match the video.

If you wish to use the libfdk_aac then I might suggest taking your result and try trimming the first 46ms off it:
ffmpeg -i "video" -ss 0.046 -vcodec copy -acodec copy "outputTrimmed.3gp"

If that fails., try trimming the last 46ms :
ffmpeg -i "video" -vcodec copy -acodec copy -t 01:31:40.81 "outputTrimmed.3gp"

It worked for me but I do not have the libfdk_aac since I do not believe its possible to get that in Zeranoes ffmpeg builds.
I hope this will help.

UPDATE: The left video is the original. Notice the audio does not come in until 0.133 or 3rd frame.:


The video on the right is with 0.134 trimmed from the beginning. This cured the late start and it was as close to the original MediaInfo as I could get from YouTube. Hope it works.


Thanks
Budman1

[hss2293]

Is there any way to prevent this delay by adding some more options in the command line ?

[Budman1]

There is probably yes but there are many here in this forum that are more knowledgeable about this than I am. All I can suggest is , with MY example, graphing the audio shows it does not start until the forth video frame or about 0.133 seconds (29.97FPS -> .0333 seconds/frame * 4 = .133) late so I could try adding -ss 0.133 AFTER the -i input. The only problem is, with my original video in MP4 format, it does not tell me that the video is late so how could I have known.

Again, I do not know why it doesn't but I would not know enough to change it originally because it does not tell me with media info. I assume its how different codecs work/decodec/package/??? You could try you original with -ss added:

ffmpeg -i filename -ss 0.046 -vcodec mpeg4 -b:v 170k -s 320x240 -acodec libfdk_aac -b:a 32k -ar 44100 -ac 1 output.3gp

By adding -ss 0.xxx AFTER the -I video it should make the seek more accurate (https://trac.ffmpeg.org/wiki/Seeking%20with%20FFmpeg) but you would only know that (if it even works) by graphing the audio or making note of the delay from the CONVERTED 3GP.

NOTE: .046 is about 21.74 FPS and since I doubt that is your frame rate on the original so where you would end up is probably more likely 2 X Frame Rate. With my example I actually ended up around 150MS seeking to 133ms as shown below. Personally I would just leave it if it truly is in sync and plays fine otherwise it will be a trial and error method unless someone has more information or your original video.


Seeking to 0.133 seconds


original, locating start of Seek result above

[hss2293]

I want to ask one more thing.. When i convert video using ffmpeg into 3gp it is being converted into 3GPP Media Release 4..
How i can convert it using 3GPP Media Release 5 ???

[Budman1]

Can't help there... I DO know that if you convert it to h264/avc (libx264) it will become 3GPP Media Release 6.

ffmpeg -i filename -vcodec libx264 -b:v 170k -s 320x240 -acodec libvo_aacenc -b:a 32k -ar 44100 -ac 1 output.3gp

That is supposed to be a valid codec for 3gp but whether it will play ok, I do not know. sorry.
From WIKI:

The 3GP file format stores video streams as MPEG-4 Part 2 or H.263 or MPEG-4 Part 10 (AVC/H.264), and audio streams as AMR-NB, AMR-WB, AMR-WB+, AAC-LC, HE-AAC v1

[hss2293]

But it will give me h264 codec profile.. Here is the mediainfo of 3gp file using 3gpp media release 5

Format                                   : MPEG-4
Format profile                           : 3GPP Media Release 5
Codec ID                                 : 3gp5
File size                                : 67.6 MiB
Duration                                 : 47mn 1s
Overall bit rate mode                    : Variable
Overall bit rate                         : 201 Kbps
Encoded date                             : UTC 2013-12-21 18:11:58
Tagged date                              : UTC 2013-12-21 18:11:58

Video
ID                                       : 1
Format                                   : MPEG-4 Visual
Format profile                           : Simple@L1
Format settings, BVOP                    : No
Format settings, QPel                    : No
Format settings, GMC                     : No warppoints
Format settings, Matrix                  : Default (H.263)
Codec ID                                 : 20
Duration                                 : 47mn 1s
Bit rate mode                            : Variable
Bit rate                                 : 170 Kbps
Maximum bit rate                         : 490 Kbps
Width                                    : 320 pixels
Height                                   : 240 pixels
Display aspect ratio                     : 4:3
Frame rate mode                          : Constant
Frame rate                               : 29.970 fps
Color space                              : YUV
Chroma subsampling                       : 4:2:0
Bit depth                                : 8 bits
Scan type                                : Progressive
Compression mode                         : Lossy
Bits/(Pixel*Frame)                       : 0.074
Stream size                              : 56.2 MiB (83%)
Writing library                          : Lavc55.12.100
Encoded date                             : UTC 2013-12-21 18:11:55
Tagged date                              : UTC 2013-12-21 18:12:01

Audio
ID                                       : 2
Format                                   : AAC
Format/Info                              : Advanced Audio Codec
Format profile                           : HE-AACv2 / HE-AAC / LC
Codec ID                                 : 40
Duration                                 : 47mn 1s
Bit rate mode                            : Variable
Bit rate                                 : 32.0 Kbps
Maximum bit rate                         : 46.2 Kbps
Channel(s)                               : 2 channels / 1 channel / 1 channel
Channel positions                        : Front: L R / Front: C / Front: C
Sampling rate                            : 44.1 KHz / 44.1 KHz / 22.05 KHz
Compression mode                         : Lossy
Stream size                              : 10.8 MiB (16%)
Encoded date                             : UTC 2013-12-21 18:11:56
Tagged date                              : UTC 2013-12-21 18:12:01

[Budman1]

No, the codec ID will be AVC1 but the codec PROFILE will be 3gp Media Release 6 in an MP4 Format.

General
Complete name : C:\Users\Bud\Desktop\outputTrimmed.3gp
Format : MPEG-4
Format profile : 3GPP Media Release 6 Basic
Codec ID : 3gp6
File size : 12.4 MiB
Duration : 2mn 35s
Overall bit rate : 671 Kbps
Movie name : New York City: A Few Clips
Movie name/More : Captured with Replay Media Catcher 5 - www.applian.com
Video
ID : 1
Format : AVC
Format/Info : Advanced Video Codec
Format profile : Baseline@L3.0
Format settings, CABAC : No
Format settings, ReFrames : 1 frame
Format settings, GOP : M=1, N=60
Codec ID : avc1
Codec ID/Info : Advanced Video Coding
Duration : 2mn 35s
Bit rate : 568 Kbps
Width : 640 pixels
Height : 360 pixels
Display aspect ratio : 16:9
Frame rate mode : Constant
Frame rate : 29.970 fps
Color space : YUV
Chroma subsampling : 4:2:0
Bit depth : 8 bits
Scan type : Progressive
Bits/(Pixel*Frame) : 0.082
Stream size : 10.5 MiB (85%)
Audio
ID : 2
Format : AAC
Format/Info : Advanced Audio Codec
Format profile : LC
Codec ID : 40
Duration : 2mn 35s
Bit rate mode : Constant
Bit rate : 96.0 Kbps
Channel(s) : 2 channels
Channel positions : Front: L R
Sampling rate : 44.1 KHz
Compression mode : Lossy
Stream size : 1.78 MiB (14%)

AS I say, 3GP is unfamiliar ground for me, so any experts please feel free to add information.

[hss2293]

No one is replying to this thread except you.. I think this forum is made for helping but no one having knowledge is replying here.

[Selur]

No one is replying to this thread except you.. I think this forum is made for helping but no one having knowledge is replying here.

a. a lot of people do not like to talk about delay&co unless there is a sample
b. most people to not care about 3gp since they prefer mp4; so when you speak of '3GPP Media Release 5' you might have to add more details on what you exactly want so people which do not use 3gp at all can help
c. no need to answer, since Budman1 does a good job answering the questions.
d. people to have a real life and family to care about (especially over holidays)

The main questions so far were:

• "Why does ffmpeg add a delay?"
  This normally happens to compensate:
  • Audio&Video delays might get added to compensate decoding problems. In example if the input video does start with only a part of a gop.
  • New Audio&Video delays necessary due to format changes different audio formats for example have different frame/block sizes which results in different delays. Also different audio encoders create different audio delays.
  • delay might be added due to container restrictions regarding some profiles/formats.
  • delay might be added due to a bug (so posting which tool versions were used isn't normally a bad idea)
  
  This was answered by Budman1, poisendeathray and myself
• "Is there any way to fix it ?"
  DB83 answered here, that there really isn't an issue to fix and Budman1 replied that there might be ways to 'fix' this delay.
  Both replied that more infos are required.
  Since you didn't provide a sample Budman1 then started to use a random file from Google and posted his experience with it.
• "Is there any way to prevent this delay by adding some more options in the command line ?"
  Budman1 posted a way and for positive delays one could also use something like http://superuser.com/questions/579008/add-1-second-of-silence-to-audio-through-ffmpeg
• "How i can convert it using 3GPP Media Release 5 ??? "
  Budman1 replied that he doesn't know since the doesn't use 3gp. Same goes for me and like mentioned above, provide more detailed infos what you want/need so others, which do not know about '3gb standard' and do not have the time to read up on it using http://www.3gpp.org/DynaReport/21101.htm can help.

-> only question that isn't answered was the last question about '3GPP Media Release 5', to figure out which settings are required the easiest way is probably to find a file which is '3GPP Media Release 5'-compatible and create content which mirrors the basic characteristics of that content.

It worked for me but I do not have the libfdk_aac since I do not believe its possible to get that in Zeranoes ffmpeg builds.

libfdk_aac and some other extensions for ffmpeg are not allowed to be distributed publicly due to license issues.
-> only legal way to get ffmpeg with libfdk aac would be to build ffmpeg yourself, in example using scripts like https://github.com/rdp/ffmpeg-windows-build-helpers

Cu Selur

[hss2293]

Here is the sample of the 3gp file which is using 3GPP Media Release 5 and mediainfo of that file is provided above ..
Can anyone now tell me how to do it using ffmpeg

[Selur]

ffmpeg -i filename -ss 0.046 -vcodec mpeg4 -b:v 170k -s 320x240 -acodec libfdk_aac -b:a 32k -ar 44100 -ac 1 output.3gp

creates '3GPP Media Release 4' content according to MediaInfo
and

ffmpeg -i filename -vcodec libx264 -b:v 170k -s 320x240 -acodec libvo_aacenc -b:a 32k -ar 44100 -ac 1 output.3gp

creates '3GPP Media Release 6 Basic' content according to MediaInfo
(even with '-profile:v baseline -level 1' MediaInfo still indicates '6 Basic', so like Budman1 thought H.264 seems to always result in '3GPP Media Release 6 Basic'.

-> no clue what requirement is triggering '3GPP Media Release 5'
since:

Format                                   : MPEG-4 Visual Format
profile                           : Simple@L1

already is a relatively hard restriction.
------------------
correction:

ffmpeg -y -i filename -ss 0.046 -vcodec mpeg4 -b:v 170k -s 320x240 -g 250 -qscale 4 -mbd rd -trellis 1 -subcmp satd -strict -2 -acodec aac -b:a 48k -ar 44100 -ac 1 h:\Output\output.3gp

produces: '3GPP Media Release 5' read the ffmpeg documentation regarding the meaning of the settings
(found this by simply googling for 'ffmpeg 3GPP Media Release 5' and lookign at: http://geekdeck.wordpress.com/2010/01/17/encoding-video-for-cell-phones/
so a simple google search could have solved that as timely as you wanted,...

[hss2293]

I have already tried this.. The above command creates 3GPP Media Release 4 file not 5. Here is the mediainfo of file converted by the above said command

Format                                   : MPEG-4
Format profile                           : 3GPP Media Release 4
Codec ID                                 : 3gp4
File size                                : 12.9 MiB
Duration                                 : 4mn 21s
Overall bit rate                         : 414 Kbps

Video
ID                                       : 1
Format                                   : MPEG-4 Visual
Format profile                           : Simple@L1
Format settings, BVOP                    : No
Format settings, QPel                    : No
Format settings, GMC                     : No warppoints
Format settings, Matrix                  : Default (H.263)
Codec ID                                 : 20
Duration                                 : 4mn 21s
Bit rate mode                            : Constant
Bit rate                                 : 361 Kbps
Width                                    : 320 pixels
Height                                   : 240 pixels
Display aspect ratio                     : 4:3
Frame rate mode                          : Constant
Frame rate                               : 24.000 fps
Color space                              : YUV
Chroma subsampling                       : 4:2:0
Bit depth                                : 8 bits
Scan type                                : Progressive
Compression mode                         : Lossy
Bits/(Pixel*Frame)                       : 0.196
Stream size                              : 11.3 MiB (87%)
Writing library                          : Lavc55.41.101

Audio
ID                                       : 2
Format                                   : AAC
Format/Info                              : Advanced Audio Codec
Format profile                           : LC
Codec ID                                 : 40
Duration                                 : 4mn 21s
Bit rate mode                            : Constant
Bit rate                                 : 48.0 Kbps
Channel(s)                               : 2 channels
Original Channel count                   : 1 channel
Channel positions                        : Front: C
Sampling rate                            : 44.1 KHz
Compression mode                         : Lossy
Stream size                              : 1.50 MiB (12%)

[Selur]

worked for me on one file,.. -> no clue what is needed to require "3GPP Media Release 5" instead of "3GPP Media Release 4"
Since you do not seem to know that either, you will probably either have to wait until someone who uses 3gp comes along.

quickly reading over http://ffmpeg.org/pipermail/ffmpeg-devel/2008-June/049902.html I really don't see a reason to aim for "3GPP Media Release 5" -> you probably will have to read through the specification to figure out which requirement triggers the usage of release 5 over release 4. (there might be some users using 3gp in the ffmpeg irc channel)

[hss2293]

I will wait for other replies and i will search also.. Thanks for the help

Channel(s): 2 channels / 1 channel / 1  channel 
Channel positions: Front: L R / Front: C  / Front: C
 Sampling rate : 44.1 KHz / 44.1  KHz / 22.05 KHz

What does this mean 2 channels / 1 channel / 1 channel and same for channel position in above mediainfo of a file.

[Selur]

running MediaInfo with full detail might show some additional infos, other than that:

Channel(s): 2 channels / 1 channel / 1  channel
Channel positions: Front: L R / Front: C  / Front: C

seems like a bug in MediaInfo, since the channel positions does not make any sense the way it is displayer atm.
-> if you want/need it fixed report it to the MediaInfo bug tracker and with a bit of luck the author deems the problem to be serve enough to fix it.

[hss2293]

Any good alternative to ffmpeg for 3gp encoding ??

[Selur]

The problem is not ffmpeg but the missing knowledge about what are the requirements for "3GPP Media Release 5" content,....
-> using another tool will not change anything unless the tool specifically has some preset for this.

[poisondeathray]

If you remux it with yamb 2.1.0.0 beta 2, it will be 3GPP Media Release 5 . So probably there is an option in mp4box to do this , or that version just happened to use 3GPP Media Release 5 specs. You can look at the mp4box documentation for more info

Or you might search for an older ffmpeg build that uses Release 5 specs if new ones tend to use Release 6. The article selur linked to was published in 2010, so look around that period . I don't think ffmpeg has a switch to control what profile is used for 3gp . (note the mp4box from yamb 2.1.0.0 beta 2 was from 2009, so maybe around that period various muxers conformed to Release 5 specs)

[Budman1]

Selur and PoisonDeathRay, Thanks for the great information (as always from both of you) and enough to keep ME reading for a while. The information and links have been most helpful to try to increase my knowledge. Appreciate the help!