A goodly chunk of the music I listen to does sound better when encoded with Vorbis RC3 than with any of the mp3 encoders I've tried.
About half of what I've encoded thusly is some variation on classical (primarily strings), soft new age and electronica, and some of the more meditative ethnic musics (such as Ravi Shankar's sitar playing). Generally speaking, at identical bitrates (128k) these are more heavily artifacted under mp3 than Vorbis. In fact, I have not noticed any significant artifacting of this type of music at the default bitrates under Vorbis, whereas it is typically painful to listen to much of this encoded similarly under one of the mp3 encoders. For these, it's a technical decision (I would compromise on the licensing and patent issues, if it were necessary, to avoid having either insanely large files or having painful artifacting).
The other half of what I've encoded thusly (rock, pop, movie soundtracks, "fuller" symphonic music) does not have noticable artifacting in either case. For those, it's the convenience of smaller files at a comparable bitrate, plus consistency of format. It's just convenient that the license and patent issues resolve so nicely.
I am aware that in certain circumstances (specifally, when built with gcc < 2.95 and some versions of egcs), the vorbis encoder artifacts horribly. This is the result of incorrect math code being output by the compiler.
In addition, converting from another lossy format (e.g. mp3) straight into vorbis typically results in severe artifacting. You need to either leave it in the first lossy format (e.g. mp3) or you need to re-encode from a pristine (non-lossy) source, such as the original CD. I've tried it, and unless the original MP3 is pretty close to CD quality, it's a really bad idea to re-encode as an Ogg Vorbis file. (This, plus the compiler issue, may be the primary contributors to the belief that Vorbis is heavily artifacted.)
[ Parent ]