[vorbis-dev] Streaming Considerations.

Monty xiphmont at xiph.org
Mon Jul 31 11:13:03 PDT 2000



------- Forwarded Message

I'm replying to two messages.

First :
Adam Scriven (scriven at lore.com) wrote :

>    I was thinking about this streaming thing again (thankfully, not about the
>    info stream, but the audio stream. 8-) ), and I was wondering if it would
>    be streamable like MP3, or more like Real Audio. MP3 players, like WinAMP,
>    can stream without a special server. This may not be as reliable,
>    streaming over HTTP, but it is quite handy. Will this be possible with
>    Vorbis (dependent on the plug-ins)? And, if it IS possible, it should also
>    pre-buffer the next coming song, if possible, so that the audio stream
>    pauses as little as possible between songs, similar to how WinAMP does it.     

I don't see any reason why OGG-vorbis  could not be streamed over HTTP
or any other TCP based protocol.
I streamed mp2 over FTP ages ago and it worked just fine.

Second :
Peter Shore (peteshore at home.com) wrote :
>    Having work extensively with Windows Media and its player control for
>    embedded objects, I was wondering very much the same thing.
> 
>    Something many people don't realize is that even using just the http
>    protocol, there are radical differences in streamability between file
>    formats. Compared to ASF audio/video, MPEG audio/video streams poorly
>    and unreliably. The former is not cached at all on a hard drive, but
>    downloaded direct to memory - usually with just enough data to maintain
>    a 5-second buffer. MPEG audio/video, on the other hand, is read only
>    from a hard drive cache.

If this happens, then you have a bad or misconfigured mp3 player or www
browser.

You can stream MP3s over HTTP with WinAMP with no problems. A 10 hour
mp3
starts playing after a few seconds ( when enough data is downloaded to
fill
the buffer ) and then plays nicely to the end.

>    In terms of practical consequence, at modem bitrates a large MP3 encoded
>    at 20k can take a full minute to begin playing via any protocol, whereas
>    a 20k ASF/WMA file will begin streaming within seconds. At broadband
>    bitrates, a 160k ASF will stream much more reliably than a 128k MP3 -
>    and is far less prone to interruptions due to temporary bandwidth
>    drop-off, and to flutter due to competition for computer bus resources.

You are probably seeing the difference between an UDP based transport
and a 
TCP based one. ( TCP sucks for radio type transmissions/broadcasts ).
 
>    So I'm hoping that OGG, unlike MP3, is a non-cached file type like ASF
>    (& which can also be a strong security feature for artists wishing to
>    stream original content).

Again , this is not dependent on the file format , but the used
transport
method.
 
>    In the long term, streamability plus adaptability to multichannel audio
>    (beyond just 2-channels), are what I think will be the most important
>    factors in the success of any format. An embeddable player object and
>    control (or compatibility with an existing audio/video player like Real
>    or MPlayer), I think are also very critical.
> 
>    Haven't seen much discussion addressing any of that.
>       
>    Regards,
>    - Peter

P.S.: I am not subscribed to the list , so this mail may not appear on
it,
unless Monty blesses it ( hint-hint :-)
- --
David Balazic
from http://surf.to/stein

------- End of Forwarded Message

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/



More information about the Vorbis-dev mailing list