opus - Aug 2015 - Feeding decoder with reordered packets

Sorry for probably obvious question but I could not find and answer on mail
archive.

What would happen to decoded audio quality if the packets fed to decoder are out
of order (no jitter buffer)?

For example, if order of packets is 0,1,2,4,3,5 (3 and 4 reordered) and there is
no FEC.
Would it be better, in this case, to drop packet 3 and not decode it at all?

Or, if there is FEC, then try to restore 3 from 4 (which is 3's next packet
and carries FEC for 3) and then drop 3 and continue to 5?

Thanks,
Pavel
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://lists.xiph.org/pipermail/opus/attachments/20150831/29e80583/attachment.htm

Hi Pavel,

The Opus decoder has no way to know that the packets you're feeding it
are out of order. If you cannot reorder them, then you want to feed in
as many as you can in order, and then use the PLC (pass NULL for the
packet) to fill the holes. In your example, it would mean passing:
0,1,2,NULL,4,5

	Jean-Marc

On 08/31/2015 03:12 AM, Pavel Punsky wrote:
> Sorry for probably obvious question but I could not find and answer on
> mail archive.
> 
>  
> 
> What would happen to decoded audio quality if the packets fed to decoder
> are out of order (no jitter buffer)?
> 
>  
> 
> For example, if order of packets is 0,1,2,4,3,5 (3 and 4 reordered) and
> there is no FEC.
> 
> Would it be better, in this case, to drop packet 3 and not decode it at
all?
> 
>  
> 
> Or, if there is FEC, then try to restore 3 from 4 (which is 3?s next
> packet and carries FEC for 3) and then drop 3 and continue to 5?
> 
>  
> 
> Thanks,
> 
> Pavel
> 
> 
> 
> _______________________________________________
> opus mailing list
> opus at xiph.org
> http://lists.xiph.org/mailman/listinfo/opus
>