streaming audio

brueckd at tbye.com brueckd at tbye.com
Mon May 6 12:19:47 EDT 2002


On Tue, 7 May 2002, Anthony Baxter wrote:

> 
> >>> brueckd at tbye.com wrote
> > If possible, avoid actual streaming protocols as they are quite convoluted 
> > and proprietary. 
> 
> Again, I point to RTP. 

Ok, just convoluted then. ;-) Seriously though, it is *far* easier to 
implement a music server by basing it on HTTP than RTP/RTSP, so unless you 
have need of specific features from those protocols, avoid the extra 
complexity and make a simple HTTP server (I'd argue that RTP/RTSP is 
overkill for what the OP wanted).

> And most applications (RealPlayer, Quicktime, others) use RTSP as the 
> control protocol for RTP-based applications.

I guess I have a rather low opinion of most streaming applications.  
Although I don't think HTTP is the best solution to the rich Internet
media problem, the current alternatives seem more like short-term hacks.
If you take a strictly *client-side perspective*, you really only need a
streaming protocol if the rate at which you consume data is greater than
your average connection speed; otherwise you can get by with an ongoing
HTTP or FTP download. My slow DSL connection far exceeds the rate at which
I consume mp3 audio data, and it could easily keep up with a mid-size Divx
video "stream" over HTTP; over a LAN I could have a separate video feed
going to many TVs.

> Ok, so last time I looked Real were abusing the hell out of RTSP and
> doing stuff in a non-compliant way,

Yup. Don't ever become involved in writing a proxy/cache for any of those 
guys; it sucks your will to live.

In the "olden days" I had to use streaming protocols for both audio and
video. My connection speed has increased enough that I *never* use true
streaming for audio, and only occasionally for video, and even then the 
quality (tiny size, loading delay, stops and jitters) leaves me wishing 
for a vanilla HTTP download with a few second pause at the start to let 
the player buffer enough of a head start.

-Dave






More information about the Python-list mailing list