Weekly Edition
Return to the Press page
| |||||||||||||
|
| |||||||||||||
What is Ogg Vorbis?
The Ogg Vorbis site has posted
an introductory document
that describes the Ogg Vorbis audio compression system.
(Log in to post comments)
Is there a specification document for Ogg Vorbis? Posted Jun 4, 2002 23:08 UTC (Tue) by DeletedUser1286 ((unknown), #1286) [Link] There is a reference implementation for Ogg Vorbis, but is there a formal specification that one could use to write an independent implementation if needed?
Is there a specification document for Ogg Vorbis? Posted Jun 5, 2002 0:24 UTC (Wed) by DeletedUser1291 ((unknown), #1291) [Link] A complete and detailed specification is what is preventing Ogg Vorbis from reaching version 1.0 right now. It is my understanding that the Xiph Foundation is well aware of this problem, and is working on these documents as fast as they can.The Ogg part, which describes the file format, has been completed for some time now. You can look at the current state of the documentation at http://www.xiph.org/ogg/vorbis/docs.html. They have not been updated in a year, though, so you probably better go out and try the technology itself. It is amazing.
No, there isn't. Posted Jun 5, 2002 0:53 UTC (Wed) by Baylink (guest, #755) [Link] And the lack of one has been the subject of quite some discussion over at Kuro5hin, referenced here a bit ago, IIRC.Their answer is, roughly, "we're pedaling as fast as we can, give us a bit..."... and you can interpret that verb howmever your like. :-) I, personally, am inclined to give them some time, but maybe that's just me.
Vorbis Propaganda Posted Jun 4, 2002 23:41 UTC (Tue) by cypherpunks (guest, #1288) [Link] I'm sorry but the Ogg Vorbis guys make a lot of unsubstantiated claims without any proper studies to back their claims. The one or two studies I do know, all claim that Vorbis quality is inferior to the best MP3 encodings.
Vorbis Propaganda Posted Jun 5, 2002 0:41 UTC (Wed) by DeletedUser1291 ((unknown), #1291) [Link] There has been no real quality test done to compare state-of-the-art Ogg Vorbis encoding (version 1.0 RC3) with state-of-the-art MP3 encoders. We have seen amateurish and unconclusive attempts, though, or tests done on old versions of Ogg Vorbis, or tests done improperly, for example using constant-bitrate settings, or even worse, tests done on Ogg Vorbis re-encodings of MP3s¹.While this is not a proper quality test, I can assure you that my ears, which listen almost exclusively to Ogg Vorbis files and Audio CDs since the beginning of this year, feel that the quality of Ogg Vorbis is superior to the quality of MP3. To be more explicit, I encode my files at quality 3.14, which generally gives me files around 110 kbps, and I think, err, I feel, I know, they sound better than the 128 kbps MP3s I have encoded in the past using the highest quality setings of the Fraunhofer encoder, and much better than the MP3s you can download from the Internet. ¹ Re-encoding an MP3 into an Ogg Vorbis is like doing a photocopy of a fax: a bad idea.
Vorbis Propaganda Posted Jun 5, 2002 1:03 UTC (Wed) by Strike (guest, #861) [Link] Speaking of propaganda, where are your findings? After several side-by-side comparisons of my OWN, I have concluded definitively that ogg is by far superior in both quality and compression. A 112kbps (VBR) ogg (quality setting of 3) sounds as good or better than 192kbps and up MP3s in any encoder I tried (bladeenc and lame), and takes approx 60% as much space. Any studies that use any ogg encoder before RC3 should be ignored, as RC3 represents a HUGE step in encoding quality. Also, RC3 has been out long enough for it to be considered the de facto standard. My only gripe with ogg so far is that since the encoder is still in RC stage, it is quite slow. 1.0x encoding rates and up are rare.
Vorbis Propaganda Posted Jun 5, 2002 9:09 UTC (Wed) by DeletedUser717 ((unknown), #717) [Link] Huh? I get 4-5x when encoding 48KHz files with -q6 (1.3GHz Athlon) This is using an Intel Compiler compile.
Speed Posted Jun 5, 2002 10:40 UTC (Wed) by DeletedUser717 ((unknown), #717) [Link]
Input: radio_0605-cut.wav 464,437,284 bytes (16bit/48KHz/stereo)
Done encoding file "radio_0605-cut.ogg"
File length: 40m 18,0s
Elapsed time: 7m 28,0s
Rate: 5,3994
Average bitrate: 119,0 kb/s
Output: radio_0605-cut.ogg 35,979,176 bytes For reference, my: LAME version 3.92 MMX (http://www.mp3dev.org/) CPU features: i387, MMX (ASM used), 3DNow! (ASM used) Manages a 4.4x encode with 'lame -v radio_0605-cut.wav'
Speed Posted Jun 5, 2002 11:22 UTC (Wed) by Strike (guest, #861) [Link] That's all good and well, but it's still slow for me :) On a P3-450 anyway.
Vorbis Propaganda Posted Jun 5, 2002 7:28 UTC (Wed) by tekNico (guest, #22) [Link] > I'm sorry but the Ogg Vorbis guys make a lot of unsubstantiated > claims without any proper studies to back their claims. So I'm going to make some other unsubstantiated claims myself. :^) > The one or two studies I do know, all claim that Vorbis quality is > inferior to the best MP3 encodings. My own experience says that such "studies" are grossly mistaken. I encoded many styles of music at many quality settings and, to quote ESR (Why Python?): "To say I was astonished would have been positively wallowing in understatement." I was hard pressed to find any difference with the original even when encoding at a quality level of 1, giving a bitrate of about 64kbs! The default is q=3, giving about 95-100kbs . To err on the safe side, I encode at q=4.99, about 125-130kbs, which is a sweet spot, since from q=5 and upwards the stereo information is coded in a lossless way, producing significantly higher bitrates. So, my conclusion is that Vorbis has the best quality/size ratio, and is free to boot! What else could one ask for?
Vorbis Propaganda Posted Jun 5, 2002 9:16 UTC (Wed) by DeletedUser1322 ((unknown), #1322) [Link] We need an objective comparison of MP3 and Ogg, were the signal to noise ratio is measured for two equally large files. This is probably the most neutral test that can be done.
Vorbis Propaganda Posted Jun 5, 2002 14:11 UTC (Wed) by DeletedUser1291 ((unknown), #1291) [Link] The only tests I know of are blind tests, where people randomly listen to encoded and unencoded sounds, and try to say which is which. With a lot of samples and a well conducted testing (which has not happened with Vorbis yet), you should be able to say, e.g. "98% of Vorbis users can't make a difference between a CD recording and a quality 3 encoded Vorbis file", as opposed to, e.g. "84% of MP3 users can't make a difference between a CD recording and a 128 kbps Lame-encoded MP3 file." The figures I give are of course only wishful thinking.
Vorbis Propaganda (personal experience) Posted Jun 5, 2002 16:38 UTC (Wed) by msphil (guest, #1378) [Link] 1.0rc3 at the default quality settings are perfectly suitable for my personal audio needs. For LAME and BLADE to reach a similar quality level (for the music to which I typically listen), 192K is a minimum rate, 224K seems to make the best compromise, and 256K is finally free of the artifacts for even the quietest music to finally sound right.(I should explain that I typically listen to symphonic music, new age(ish), and soundtracks, which are often butchered by most mp3-based approaches. I find 128K mp3s from the above to be suitable for very "noisy" music, e.g. my MC Hawking collection.)
Low Quality Settings Posted Jun 5, 2002 0:46 UTC (Wed) by DeletedUser1291 ((unknown), #1291) [Link] This is what impressed me the most when I first tried Ogg Vorbis at the beginning of the year: it manages to sound very good at quality 0, which gives you files around 58 kbps! Yes, you can easily notice some artifacts, but even them sound better than the MP3 artifacts.Be sure to test Vorbis at quality 0, you'll be impressed.
|
|
||||||||||||