HTTP and MP3: The Only Streaming Media Standards That Matter
Posted 29 Oct 2001 at 05:29 UTC by robla
If one looks into the topic of streaming media standards, one quickly
runs into confusion and hype. Lots of people will tout the virtues of
one acronym or another. I personally make a living pelting people with
acronyms and coming up with new ones. At the end of the day, though, the
only standards that matter right now are HTTP and MP3.
This may not be terribly surprising to those that frequent Advogato, but
it's something that it pains me to say. I'm one of the co-editors of
the RTSP specification, and have been
a very active member of the W3C's
SYMM Working Group working on SMIL.
The headline and teaser above is to a much longer
article that I wrote for a new streaming media community site called
Übergid. I encourage you to
post your comments over there, as well as here, since as that community
builds, there will be people who are not necessarily part of the open
source community. Hence, you won't necessarily be preaching to the
choir if you talk open source there.
Speaking of open source, a couple of comments did come in about Vorbis.
Personally, I've been a big fan of Vorbis, egging on Jack Moffitt
(jack) to do the RealPlayer plugin (for which an alpha
quality plugin is checked into their CVS repository). I wish the
project well, but that's a great example of "nerds flogging an agenda"
which I refer to in the Übergid article. The fact of the matter is
that the Thomson folks have managed to walk the tightrope pretty well,
and though the more savvy open source people may despise the success of
MP3, there's enough MP3 content that it'll always be a standard (and
hey, patents do expire).
At any rate, I do have some specific issues that I'd like to cover with
the open source community. Right now, the heir apparent as a streaming
media standard is MPEG-4. However, as has been covered here on
Advogato before, the MPEG-4 specification is laden with patents, so
much so that the M4IF can't even agree on a licensing policy for those
who are willing to pay.
I think there is an opportunity to ensure that a less hideously
encumbered video standard becomes the content winner. As near as I
know, H.263 is clean
(no one has asserted any royalties on this specification). Given that
MPEG-4 video is based on H.263, and that H.263v2 (the 1998 version) is
just as good, it seems silly to rely on the encumbered version. Both
Apple Quicktime and the RealOne Player support this today.
However, there's a dearth of open source software that implements this,
and even when it does, it doesn't support enough other standards (RTSP,
RTP, MP3) to make it compelling. One project I hold out a lot of hope
for is the Live.com Streaming
Kit, which includes support for all of the above. However, it's an
engine that needs to be incorporated into some other project.
Note that on Windows, it's pretty easy to create a file that has MP3
audio and H.263 video. Download VirtualDub and the Intel
H.263 codec, and then convert an AVI file to use the "I263" codec.
Alternately, you can download Quicktime, and use it to create your file.
However, just by putting it in a Quicktime file, you may have created
I'd like to get some thoughts, and I hope that I inspire some activity
on this front.
QoS transport are doomed. Until people are ready to pay for a
distinctive different QoS (and this usually doesn't come cheap at all)
all efforts toward builting such transport are doomed. So far the
attitude "if there is enough people interested, the pipe will grow,
and latencies will stabilize" has worked, it might get harder due to
the negative economic environement, but so far the improvement is
But it is important to have formats not encumbered with Patents. I
applaud the work of the Vorbis group, maybe there is a scope problem
in their effort (hard to not fall in that pit !), but it's very
important to have at least a fallback strategy if the patent holders
starts to get annoying.
Aside, so when are we gonna get OpenSource SMIL2 players, Nabil
has some code too, what's the current state ;-) ?
Yes!, posted 29 Oct 2001 at 20:52 UTC by Bram »
Standards bodies exist to provide
sage wisdom which implementers actually benefit from following.
'Standards' must have the following characteristics -
- Clearly necessary
- Not have a reasonable existing alternative
- Easy to understand and implement
- Not patent encumbered
If a 'standard' doesn't conform to all of the above, it's won't get
implemented and will just add confusion. We're better off with no
standards at all.
VP3?, posted 30 Oct 2001 at 04:34 UTC by wmf »
I'm glad to see someone else beating the H.263 drum. Hearing people complain that there's no standard, open source, low-bandwidth video codec gets old when it's not true.
What do you think about VP3? Does it have a chance? (Besides the fact that they need to get some clue and remove the registration requirement from their site...)
Multicast, posted 30 Oct 2001 at 17:56 UTC by jwb »
What is the sad story of multicast? It seems that no significant
streaming ever takes place over the multicast net, yet multicast is a
clever and elegant solution to the streaming bandwidth problem. The
head-end needs only as much bandwidth as each client.
I understand that in the past, Cisco equipment was mostly incapable of
understanding multicast traffic, but that isn't the case anymore and I
wonder why multicast still lacks momentum.
Multicast creates at least one route entry per group in every router that the group's traffic passes though. I get the impression that Internet core routers don't have enough spare memory for that.
right now..., posted 31 Oct 2001 at 02:58 UTC by graydon »
I do not think it is MP3 nor HTTP per se which are the issue.
people are now accustomed to an interaction mode ("content" hosted on or
streamed from "sites") which is unlikely to change. but it will be
relatively easy to slip from MP3 to OGG (or, unfortunately, WMF), just
or HTTP features (ubiquitous proxying, etags, pipelining) into the
browser, or new mail features (CRAM-MD5, ESMTP, APOP, IMAP) into the
the reason is simple: there are only a handful of well-used, credible
access devices (browsers, streaming media players) and they all work
exactly the same way. when a point release comes out that supports a new
"format" that works exactly the same way the old one did, no user
expectations differ, and the user base gradually, unknowingly
transitions to the new technology. I do not see any reason to consider
waiting the years for MP3 to expire, when there are compelling reasons
to enable OGG now, both for the developer and the end-user, and the
average shoutcast.com listener or mp3.com downloader will not likely
even be aware of when they are using a vorbis (or WMF, sigh) stream,
since they did the same thing they always do (activate player, select
stream / artist of choice)
I thought people here may be interested in the discussion
going on over at the MPEG4IP forum on SourceForge. We're starting
to get an
interesting discussion going about H.263/MP3/RTSP/RTP interop. Still
early, but I think there's some definite possibilities there.
While I don't agree with the tactic (promoting MPEG-4), the spirit of
the project is a good one. They seem very interested in talking about
interoperability using whatever means possible. The thread concerns
asking them to widen the scope of things.
For a patent free environment MP3 is irrelevant and MP4 is even less
appealing. OGG streaming works. Vp3 needs a lot of work to clean up the
released code but could be good - its certainly an equal of DivX without
the patent disaster that will turn into
Don't overlook the CU30/QVix codec for real time video either
...it needs to be taken to an established standards group. You go to Xiph.org, and there's an
rant in the "about" section, and no
feeling of legitimacy. I might be able to get folks here at RealNetworks to take Ogg seriously if it were submitted to the W3C, IETF,
OASIS, ITU, or ISO (probably in that order). What would be especially good about W3C is that it would give them an opportunity to
that they are still very much about royalty-free standards (and, make no mistake, at the core, they are...it's just hard to get membership
consensus around that, and not 100% necessary in all cases)
Now, whether or not Ogg needs any corporate help for it to succeed is entirely another question (one I won't try to tackle here). However,
being someone who works at a corporation who is genuinely interested in fostering tangible interoperability, it's not a practical option at
this time, and MP3, though encumbered, is a usable solution.