In light of the recent study - http://www.airwindows.com/encoders/index.html - posted at slashdot, I've come to realize how little I know about this whole field. So, I'm making an appeal to the audio experts here. I want 'jukebox' quality, meaning ~0 artifacts while maintaining a good compression ratio, but not quite 'archival' (read high bitrate) as I don't mind hanging on to the original cds. Is r3mix's recommended "lame -V1 -b128 -mj -h" good enough? Is this guy and his study saying that it's *no where near* good enough as it seems? Is this study for real, or is it bunk? I realize that this seems like a question more suited to some mp3 mailing list, but given that the experts here have no immediately obvious attachment to any mp3 encoder or decoder, I'm hoping to get unbiased opinions. And I know that ogg is not quite there yet, but I'm a little confused as to ... Various stuff I've read on xiph.org seems to indicate that ogg/vorbis was born out of a desire for a 'free-speech' lowbitrate streaming/compression format, not for '~0-artifact digital jukeboxes'. Is this one of the goals for 1.0? When can I completely throw out mp3 and put my cds in a timecapsule? --- >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 Sun, 29 Oct 2000 14:58:30 -0800 Joe Soroka <vorbis@xiph.org> wrote:> > In light of the recent study > - http://www.airwindows.com/encoders/index.html - > posted at slashdot, I've come to realize how little I know about this whole field. > > So, I'm making an appeal to the audio experts here. I want 'jukebox' quality, > meaning ~0 artifacts while maintaining a good compression ratio, but not > quite 'archival' (read high bitrate) as I don't mind hanging on to the > original cds.Me too!> Is r3mix's recommended "lame -V1 -b128 -mj -h" good enough?In my experience, yes. These setting sound as good as 256kbit/s and way better than 128kbit/s to my tin ears. A lot of nasty artifacts that I used to get with early Fruanhofer and ISO derived codecs are nowhere to be heard, and hasn't mucked up the stereo imaging or given noticable pre-echo effects on anything I've tried.> Is this guy > and his study saying that it's *no where near* good enough as it seems? Is > this study for real, or is it bunk?Mostly bunk. The fact is that perceptual encoders are supposed to throw out information that cannot be heard, so the image generated by the encoded sample will almost necessarily be different - and it's difficult to correlate a conspicous difference in the images to encoding errors without listening to each sample. Trust your ears, basically. The graphs on r3mix are a little more useful as they will tell you whether a particular encoder is using a low pass filter to drop anything over 16kHz, or whether it does a bad job of encoding the top 16-22kHz worth of sound without one - but vorbix beta 1 flunked this test as it had a bug that was specifically triggered by this kind of frequency sweep test, but which otherwise didn't show up normally.> And I know that ogg is not quite there yet, but I'm a little confused as to ... > Various stuff I've read on xiph.org seems to indicate that ogg/vorbis was > born out of a desire for a 'free-speech' lowbitrate streaming/compression > format, not for '~0-artifact digital jukeboxes'. Is this one of the goals for 1.0? > When can I completely throw out mp3 and put my cds in a timecapsule?High bitrate encoding is in their at the moment, but I expect it will want a lot more tweaking to do archival quality at bitrates that compete with LAME. I think I heard that in theory vorbis could do 160 av VBR at archival quality (without wavelets) compared to LAME's 180 av. That's way after 1.0 is released, though. John --- >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.
Joe Soroka <oggustus@zxmail.com> wrote:> So, I'm making an appeal to the audio experts here. I want 'jukebox' > quality, meaning ~0 artifacts while maintaining a good compression > ratio, but not quite 'archival' (read high bitrate) as I don't mind > hanging on to the original cds.One thing first: there are two types of artifacts - one that can be easily heard by most people, such as pre-echo, high-frequency distortions... and the other type is so called grey-area: it is how the encoded file generally sounds - is it natural, closer to the original, sharp & brittle.. - these kind of artifacts can be pretty subjective. Currently only MP+ has the quality level you want (and myself, as well). It will give you (using 'normal' profile) avg. bitrate between 160-192kbps (most of the time) with great sound quality - no obvious artifacts, even with some very hard-to-encode test tracks, and generally it gives pretty natural sound. But it is still in development, and once it is finished it won't be free, because the author has to pay some patents (he is using some Philips filters), and of course because he wants to make money. :-)> Is r3mix's recommended "lame -V1 -b128 -mj -h" good enough? Is this guy > and his study saying that it's *no where near* good enough as it seems? > Is this study for real, or is it bunk?"lame -V1 -b128 -mj -h" really gives the best quality/compression ratio in MP3 field, but that quality is just not good enough for me - too much pre-echo for my taste, and there are still high-frequency distortions. Also, one of the reasons why I don't like MP3 in general is that it adds a small amount of silence to the begging/end of the encoded file and this is heard as a click between seamless (gapless) tracks - live albums are a good example.> Various stuff I've read on xiph.org seems to indicate that ogg/vorbis was > born out of a desire for a 'free-speech' lowbitrate streaming/compression > format, not for '~0-artifact digital jukeboxes'. Is this one of the goals > for 1.0?The option to encode in high bitrates is already there, in beta2 (beta3 coming out very soon), but there are still tweakings and bugfixes to be done before Vorbis can be used for archival purposes.> When can I completely throw out mp3 and put my cds in a timecapsule?You can wait until Vorbis matures enough. That's what I'm doing. :-) Greetings, Aleksandar --- >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.
Jack Moffitt <jack@icecast.org> wrote:> > Currently only MP+ has the quality level you want (and myself, as well). > > Beware of MP+. It is not open source. It is not intended to be free. > You might as well encode your files in EPAC or something equally > disgusting from an IP standpoint.Yep. No matter how good it sounds, when it is closed source and not free, it won't get much attention. That's why I'm sticking with Ogg Vorbis. :-) Greetings, Aleksandar --- >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.