[FFmpeg-user] Question about AAC fixed duration encoding

Osama Alshaykh osama at nxtec-corp.com
Wed Apr 5 19:57:09 EEST 2017


Hi Everyone,

This is other information we see when this issue happens:

[matroska @ 0x30c7760] Non-monotonous DTS in output stream 0:1; previous: 893551, current: 893513; changing to 893551. This may result in incorrect timestamps in the output file.

[aac @ 0x30cbc20] Queue input is backward in time


We would appreciate your help.  how can we avoid having such issues.  We are using ALSA for audio and x11grab fro video.

Thanks,
Osama


> On Apr 4, 2017, at 5:22 PM, Osama Alshaykh <osama at nxtec-corp.com> wrote:
> 
> Hi Everyone,
> 
> We use FFMPEG as our internal encoding tool.  We use a third party streaming server, where they encode the videos before streaming them.  We have noticed areas where the videos get out of sync by their encoder.  We traced the issue with them and their response back to us is:
> 
> "The product team noted that normally the ffmpeg encoder should produce audio samples of fixed durations.   But they  have not used “-acodec libfdk_aac” before.   They suggest that you can try “-acodec aac”.   In addition,  they noticed that your  input source has many audio frames with duration of “1” instead of “1024”. That implies the input source might have some overlapping or discontinuity. “
> 
> My questions are:
> IF we switch to -acodec acc, will this produce fixed lengths?
> I have researched if we can set the duration length to be fixed duration.  I am not sure I did proper job in my research, can you give us an advice of how to do that?
> 
> Our container is MKV.   We transcode it afterwards to MP4 before we upload it to them.
> 
> Thank you,
> Osama Alshaykh 
> 
> _______________________________________________
> ffmpeg-user mailing list
> ffmpeg-user at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-user
> 
> To unsubscribe, visit link above, or email
> ffmpeg-user-request at ffmpeg.org with subject "unsubscribe".
> 



More information about the ffmpeg-user mailing list