Re: [AVT] draft-herlein-speex-rtp-profile-01

[Colin Perkins <csp@csperkins.org>]

Michael,

--> "Michael A. Ramalho" <mramalho@cisco.com> writes:
>--> Colin Perkins writes:
>>- The use of the M bit in the RTP header to signify comfort noise is also
>>   unusual. This is not necessarily inappropriate, but it is important
>>   to consider if the advantages of this use outweigh the costs of being
>>   different to other audio payload formats (which use the M bit to
>>   signal the first packet after a silent period).
> 
> This may be contrary to the guidance I got from Steve a while back.
> 
> A little background first ..
> 
>  From RFC 1889:
> 
> "marker (M): 1 bit The interpretation of the marker is defined by a 
> profile. It
> is intended to allow significant events such as frame boundaries to be
> marked in the packet stream. A profile may define additional marker bits
> or specify that there is no marker bit by changing the number of bits in
> the payload type field (see Section 5.3)."
> 
> But from RFC 1890 (Audio and Video Conferences with Minimal Control
> that can be used with G.711):
> 
> "For applications which send no packets during silence, the first packet
> of a talk spurt (first packet after a silence period) is distinguished by
> setting the marker bit in the RTP data header. Applications without
> silence suppression set the bit to zero."
> 
> Then Steve said on 2/2/2003 (in regards to the RGL codec) ...
> 
> "I believe you should follow the marker bit convention as specified in
> RFC 1890. The RGL codec is likely to be used in implementations that
> also use other payload formats (such as PCMU) specified in RFC 1890,
> and the behavior should be consistent."
> 
> Thus, since I think that speex will also be used in similar
> deployments (i.e., packet voice) ... I would think Steve's
> guidance should apply here as well.
> 
> My question to the speex authors is that if they simply define
> the periods in which they desire to send comfort noise to be
> *identically* the same definition as a RFC 1890 "beginning or
> ending of a talk spurt", then there would be no inconsistency
> in the bit's meaning.

I'm not sure I understand your suggestion. I was expecting a period of
comfort noise to be treated identically to a silent period.  Hence the
M bit would be set on the first packet after the comfort noise (and not
set on any of the comfort noise packet).

-- 
Colin Perkins
csp@csperkins.org

_______________________________________________
Audio/Video Transport Working Group
avt@ietf.org
https://www1.ietf.org/mailman/listinfo/avt