AES67 sender pipeline options

[Sebastian Dröge]

Hi James,

On Tue, 2019-01-29 at 19:38 -0600, James Cowdery wrote:
> I see the RTPs markers are the symptom not the cause of the discontinuities.
> I went back my code just before I added rtpbin and it doesn't have the
> regular 40ms discontinuities.

That seems unlikely to be the cause. rtpbin is not causing any changes
to the RTP packets (like setting the marker flag) and also wouldn't
cause timestamp differences. This is all happening before the rtpbin in
the payloader.

Are you sure you have the same properties on the payloader? The
"perfect-rtptime" property for example might be related to this.

> It also doesn't have the fast preroll and just starts immediately at
> 1 packet per ms.

What do you mean with this exactly?

> That code plays back OK with an 8ms fudge factor to get initial
> timestamp-offset right.

Where do these 8ms come from? That sounds like your code to set the
timestamp-offset is not correct yet. In theory you should get this
exactly correct based on the timestamp of the incoming buffer.

Your code to set the timestamp-offset should not do anything with the
clock itself, like measuring what time it is now.

> I'll need RTCP at some point but for now I've ditched it for now and work on
> getting that clean and robust.
> 
> Thanks for advice on setting the payloader timestamp offset. The method I
> have working is accurate enough for now but I'll look at these methods to
> try and get rid of the fudge factors. From reading SMPTE ST 2110 it seems
> the RTP has to be a sample of the audio clock i.e. the SDP offset should be
> 0. That matches all the streams I've seen.

I wasn't aware of that constraints of ST2110, but for e.g. AES67 or
Ravenna this is not required (or RFC7273 in general) and non-0 media
clock offsets are allowed.

-- 
Sebastian Dröge, Centricular Ltd · https://www.centricular.com

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 963 bytes
Desc: This is a digitally signed message part
URL: <https://lists.freedesktop.org/archives/gstreamer-devel/attachments/20190130/1a73f06c/attachment.sig>