Ogg player IP announced

Okay, not that exciting, but I guess it shows Ogg making inroads.

We don't need people spending time optimising a DSP architecture for Ogg playback (there just HAS to be a bit order reversal instruction in there) - what we need is for Xiph to fix up their design to be remotely usable in an embedded architecture. Right now it's a complete and utter mess. I think those guys are starting to learn why Fraunhoffer, Dolby, MS et al do certain things in certain ways.

I suppose you can't complain about a hobby project, but they're the ones who keep saying they want to be taken seriously.

Rob

Ouch.

Ouch.

Yes, but the Good Thing (TM) is that I think that there are no bunch of folks who would more like to see fellow travelers succeed than The Guys .

I may be wrong, but if I were The Guys at Xiph, I would read Rob's comments and try to process that data in as positive a fashion as possible...

Well, "mess" might be the wrong word for it. "Open ended" design is all very well when you have a big fast PC, but there are perfectly valid OGG files which could cause a gigahertz PC system to crawl due to the lack of limits on some parameters.

The problem is very acute on systems with limited memory, cache, and cpu speed.

Hugo

Yeah I didn't mean to be overly negative, I'm just stressed with expending lots of effort on a feature that there's little commercial justification for (i.e. it needs to be easier or implementors won't bother with it).

We have it running pretty well now but it sucks battery like crazy compared to MP3 and WMA. There are valid encoder settings, though, which are completely impossible to decode on a typical embedded processor (actually a relatively fast embedded processor in fact).

Rob

if I were The Guys at Xiph, I would read Rob's comments and try to process that data in as positive a fashion as possible

No, disregard my comments and we'll put together a post mortem for the Xiph guys once Pearl is shipping. I think they know the issues for themselves, though - at least some of the source comments imply that they do.

One of our guys has a partially implemented Vorbis decoder that he has been writing in his spare time, which might be more suited to embedded architectures as long as certain encoding restrictions are recognised.

Rob