I'm in the process of encoding a whole wack of CDs and found that Ogg suites me perfectly. Should I go ahead and use Garf's GT3b1 or wait for b2? Also, I've been reading around and discovered that GT3b1 with q -5 is supposedly where the stereo separation comes into play. Am I correct in thinking this?

Go ahead and use GT3b1. Garf has recently said that he doesn't have the time to work on it right now, so it could be quite a while before GT3b2 is available.


Ogg Vorbis has stereo separation throughout, but it does like to throw away stereo details that most people can't hear. Any good lossy encoder will do this.

Unfortunately, Dolby Surround (and certain other post-processing effects) rely on information in the stereo signal that ordinary people can't hear. That's why Dolby chose that method of encoding: encoded material sounds okay even to people without their decoder.

Somewhere between -q 5 and 6, Vorbis starts saving enough of the otherwise inaudible stereo image for surround encoding to survive.

Don't take my word as Gospel, but I believe starting at q 6, the stereo seperation encoding becomes lossless.

Also, although you didn't specify at what value you'll be encoding at, Garf only modified the higher bitrates, so if you were planning on using q 3, then it doesn't matter which you do (in fact the more recent builds of non-GT3 may be better for that level, since Garf hasn't updated it in a while).

Finally, the only difference I could really tell between GT3 and the Xiph's version is that GT3 is simply more liberal with the bitrates (and hence will produce a larger file for the same q setting). The full purpose of this has not been fully explained to me, except that presumably, GT3@q 10 will be of better quality than Xiph@q 10.

(On a that note, what is the most appropriate title for the non-GT3 builds)

It's something to consider when you're someone like me, who's paranoid about using beta code to archive.

Actually guys, beta 2 came out a while ago. However, its based on RC3 code, so its a "do not use" version. I asked Garf about what he has planned for beta3, and here is his response.

"Beta 3 needs 2 things:

1) adaptive way of encoding transients (I worked out the idea, and Dibrom told me APS works similarly, but no code written yet)

2) Pin down noise issue

I have no idea when I will have time for this. " - Garf
--------------------------------------------------------------------------
Beta 1 was more of an attempt to accomplish objective 2 for the q4 - 10

You mean beta2 here.

Beta1 is mostly better transient coding and some minor tweaks that probably noone will notice.



GT3 is smarter when to use more bits. This means that at the same bitrate, GT3 will have better quality. It also means that you don't have to manually crank up the quality level to get acceptable quality for difficult tracks; the encoder will figure it out itself (as was the intent of quality levels in the first place). You are right that GT3 at a quality level will almost always give bigger files than Vorbis 1.0 at a quality level. As such, it's not really fair to compare their quality that way.