I’ve read some of the topics and so far I can understand that VBR is better compared to ABR. But is it just on terms of quality or on size and speed too?
And referring to the quality, what setting of the -V switch for VBR is better than ABR? For example if I use:
VBR: -m s -q 0 -v -V 4 -b 128 and
ABR: -m s -q 0 --abr 160 -b 128 would VBR still give better results, or should I use -V 2?


Thanxxx

I suggest you use only -V 4 and drop all other switches.

I suggest you use only --alt-preset switches. It's usually better than ABR and simple VBR (-V x).
-m s is bad, -q 0 is stupid. Please search the forum and you'll find the reasons.

Here's a really obvious link: Recommended LAME settings

It is not better, it is different.
In VBR encoding is based on achieving constant quality through the stream.
In ABR encoding is based on achieving desired size, while trying to keep quality constant.
In CBR encoding is based on using constant bitrate.

Can you tell, why Stereo is bad, and why q-0 is stupid.?

It depends on whether you want high quality or fast encoding.
I personaly prefer high quality. So -q 0 might take a minute more compared to -q 2 but I don't mind at all.
Take a look here too =>

I can answer first question.
- m s As I remeber will forse Stereo mode 0, i.e. stereo type Stereo will be used.
It is less effective than mixed Joint Stereo type, so will result in quality degradation.

Interesting link. Extract:


According to the (dead) link, insane presets are --alt-preset one, whereas the command line posted above is absolutely not insane (-q 0 = real time encoding with a good 2.5 Ghz CPU with latest lame encoders : pretty sane switch ).

This site and its flaws/obvious lies has been discussed before. Read this thread and use plain -V settings if you use LAME 3.95 or higher and one of the --alt-preset's if you use an older version.

Both issues are covered extensively by other postings. A search will find a lot of very illuminating info on the subject.

Stereo is bad: Plain stereo wastes bits that an intelligent mid-side scheme (like LAME's joint stereo) could use for encoding your music. This has been discussed and argued about many times before.

q0 is stupid: q0 uses a different approach to the final losless compression (Huffman Coding) of the blocks. It doesn't increase quality but can reduce file size by a small amount (a maximum of only a couple of percent) in some cases. However it does this at a cost of about 5x the encoding time. In other words you are wasting your time for no gain.

This is mentioned in another thread, but I'll point it out here. If your primary listening is clasical, combo jazz or other light textured/soft music you will have to bump to the next quality setting or maybe even two to get uniform bit rates: Say from -V 5 to -V 4 or even -V 3.

Sometimes -V 5 or even -V 4 will end up with bit rates as low as 80 on this type of music, in my experience with a few thousand tracks. Then when you turn up the volume to hear the details, like you have to anyway on this type of music, the quality is easily distracting. In fact it is enough of a pain trying to guess what setting for each track you need to get around 128 (my goal for predictable size and battery life) I just use --preset 134 now on all classical and jazz. That preset size averages around 128 on Lame 3.96 over a long collection. Amazingly, even with --preset 134 I have had piano tracks drop down to 104 sometimes.

The members with other listening tastes can give some tips on encoding that stuff.

Firestarter trolling moved here. Have a nice and productive day.

Try "-V 5 --athlower 1 --athaa-sensitivity 1" instead of -V 4 or -V 3. Adding the athlower and athaa-sensitivity switches will help improve the quality of quiet music, at a smaller bitrate penalty (generally) than moving up the quality scale. Of course if you aren't comfortable using advanced switches then perhaps you should just stick to the simple -V switch...but if you want to optimize the size/quality trade-off, these switches can come in handy.