Dropping the other topic as requested, I'd like to follow something up in a new thread. I want to throw a question or two out there and see what responses people give (if any ;-). Assuming that you think Vorbis could have developed faster than it has done in the past year, what in your opinion has been holding it back and what remedies do you suggest? Speaking for myself, I think that it has not attracted enough enthusiastic people who come with the knowledge required to understand how to tune Vorbis. Remedy: Lower the difficulty level for tuning Vorbis. Suggestion: Create a How-To Tune Vorbis document that takes people step-by-step through the tuning process. Dealing with a flood of people who claim to have made tuning improvements would be a nice problem to have. :-) ~Mark __________________________________ Do you Yahoo!? The New Yahoo! Shopping - with improved product search http://shopping.yahoo.com --- >8 ---- List archives: http://www.xiph.org/archives/ Ogg project homepage: http://www.xiph.org/ogg/ To unsubscribe from this list, send a message to 'vorbis-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.
On Mon, 2003-09-29 at 12:24, SwiftBiscuit wrote:> Dropping the other topic as requested, I'd like to > follow something up in a new thread. I want to throw a > question or two out there and see what responses > people give (if any ;-). > > Assuming that you think Vorbis could have developed > faster than it has done in the past year, what in your > opinion has been holding it back and what remedies do > you suggest?It was mostly held back by Monty working on Theora and portable Vorbis playback. That no-one other than Monty has the detailed knowledge it takes to make major improvements to Vorbis is unfortunate, but it's nothing that Xiph can change directly.> Speaking for myself, I think that it has not attracted > enough enthusiastic people who come with the knowledge > required to understand how to tune Vorbis. > Remedy: Lower the difficulty level for tuning Vorbis. > Suggestion: Create a How-To Tune Vorbis document that > takes people step-by-step through the tuning process.That would take a ton of time, and it's not guaranteed to attract outside contributions. The source code is out in the open, and I find it sufficiently clear to follow. Also, interested parties are always welcome to ask specific questions, but they have to be specific, i.e. Monty doesn't have time to answer broad questions like "How does Vorbis's psychoacoustic model work?" Improvements by third parties do happen, even without the documentation you suggest. Garf's tuning is one example. The recent improvement to low bitrate behavior on quiet signals is another one, and I found that one with nothing other than the source code and some common sense. Best regards, Carsten Haese. <p>--- >8 ---- List archives: http://www.xiph.org/archives/ Ogg project homepage: http://www.xiph.org/ogg/ To unsubscribe from this list, send a message to 'vorbis-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.
> -----Original Message----- > From: engdev [mailto:engdev@bigpond.net.au] > Sent: 30 September 2003 07:09 > To: vorbis@xiph.org > Subject: Re: [vorbis] Why is Vorbis development slow? > > OK, OK, I get the message. I was only 'following up' with my > thoughts which > were provoked by a user wishing to write his own > implementation, only to be > told to refer to the code. This is not what he wanted, he > wanted to create > from 'scratch' if you like, from the documentation. > > This was several months back. I never saw him on this list > again - I assume he gave up and "copied" the code.I've done my own "from scratch" implementation, as have one or two others here I think. I think you're being a little unfair on the documentation. It took me only a few days about 6 months ago to get it all working, and there was only one section which was obviously in error (the residue packet decode). For that I had to look at the libvorbis source code to figure out where the pseudocode was wrong, and I only lost an hour or so over it. After that there were still differences in decoding, so I used tons of debug output from libvorbis to figure out where mine was doing something different. No surprises there - that's exactly why you get "reference" decoders. Basically, I can see very little lacking in documentation or in the quality of the reference codec. If there's a reason people aren't adopting Vorbis it's not because of that. I suggest it's probably more to do with support and testing costs, and not to mention the usual lack of time. In case anyone's interested, my decoder is at http://www.pslam.demon.co.uk - it's a lot cleaner now, but still not entirely complete. It's mostly an experiment in how light-weight I can get it. - John Ripley. --- >8 ---- List archives: http://www.xiph.org/archives/ Ogg project homepage: http://www.xiph.org/ogg/ To unsubscribe from this list, send a message to 'vorbis-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.
> From: Sebastian Gesemann [mailto:sgeseman@upb.de] > Sent: 30 September 2003 11:53 > To: vorbis@xiph.org > Subject: Re: [vorbis] Why is Vorbis development slow? > > On Tue, 30 Sep 2003, engdev wrote: > > > OK, OK, I get the message. I was only 'following up' with > my thoughts which > > were provoked by a user wishing to write his own > implementation, only to be > > told to refer to the code. This is not what he wanted, he > wanted to create > > from 'scratch' if you like, from the documentation. > > > > This was several months back. I never saw him on this list > again - I assume > > he gave up and "copied" the code. > > Might be me... > > Well. It wasn't easy, but I succeded in overcoming the difficulties. > I had to fill the information gaps in the spec with my common sense. > (It didn't work for the floor-0 stuff, anyway) > > Wow, and i did'n look at the reference code once. > So, I'm really kinda proud. :-) > > The specs aren't perfect, but with enough background knowledge about > other source coding techniques, it's possible to build a vorbis > decoder without the reference code. > > @ John Diplay :I assume that's supposed to be me :)> You didn't implement Floor-0 either, did you ?It didn't seem worth the effort, seeing as I don't have any floor-0 files anyway. I notice that Tremor has trouble running real-time on embedded (portable) players when it encounters floor-0 files. I should think it's possible to get it much faster, but the algorithm itself doesn't exactly lend itself to being efficient. I had to look at the reference code for the residue packet decode. The pseudocode was far too mangled to make head or tail of back then, and there's a lot of "ok" edge cases to handle (premature end-of-packet being "ok" really complicates matters). I think it has been corrected now. I'm now quite tempted to properly finish off my decoder and use it instead of Tremor in the portables I work on, if I could only work out the rights nightmare that would ensue... - John Ripley. --- >8 ---- List archives: http://www.xiph.org/archives/ Ogg project homepage: http://www.xiph.org/ogg/ To unsubscribe from this list, send a message to 'vorbis-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.