Is vbr always better than abr at the same average bitrate?

No, at medium to low bitrates (140 and below, i would say), ABR increasingly becomes the better choice. Talking about LAME, of course.

Yeah, Lame definitely.

then --alt-preset xxx should use vbr at 140kbps or above.

No, that's what the VBR alt-presets are for. How do you wanna control VBR bitrate like you do with ABR? And ABR can also be useful sometimes... my brother uses higher-bitrate ABR quite often, because VBR is too slow for him on his old PC.

What about 2 pass encoding? Is it possible to implement such a way like in divx encoder?

I think 2 pass encoding can make vbr at certain final avg. bitrate.

2 pass would rock hard

but simes like nobody wanst to implant this in ther encoder

I´m using a 3. party programs to get a kind off 2 pass ABR

What programs are able to encode 2 pass ABR?

what sort of data can I use from an mp3 frame header to know if the file was ABR?

is there an extra tag like the 'Xing' tag?

cheers!

It's easy to tell ABR... check the file out in Winamp or in EncSpot... the majority of the frames should be a certain bitrate (128 for example), with the immediate higher and lower bitrates (112 and 160 in this case) being the next-most used.

Typically, by comparison, VBR usually looks more evenly distributed among 2 or 3 different bitrates... and more random too. Some music will produce a lot of low or high bitrates in a VBR MP3...

I've always wondered... WHY is VBR inferior to ABR at lower bitrates? Is it just because VBR is not yet well tuned for these bitrates? Will it beat ABR in later versions of LAME?

Do you think the difference between abr and vbr is only the way it distributes bitrate?

Well, there's is a difference how the desired bitrate is chosen. Check out this message: http://www.hydrogenaudio.org/forums/showth...12347#post12347

Basically ABR uses more rough method called perceptual entropy, while VBR relies on allowed quantization noise.