I know, that there is an overlap of "3.95.x" and "always the newest", but if you intend to use always the latest version, then you should select one of these options.

3.96

If people aren't using the recommended HA version I'm interested to know why.

I use LAME 3.90.3 for my encodes. It's slower than the latest 3.95.1 and 3.96 builds but I don't want to sacrifice quality for speed. I get 2-3x on my AMD Athlon 800MHz and 6-8x on my Pentium 4 2.8GHz with 3.90.3 so its really not that bad.

3.96

3.96
Did some ABXing and it sounds identical to 3.90 to me with the standard preset. 3.96 being a lot faster at encoding.

edit: clarification

slightly modified CVS

Because I trust the developers and what they do. And so I belive, that newer versions are in the majority of encoded samples better and/or smaller and/or faster encoded than encoded with older versions.

I don't belive in the myth of "most tested" version of 3.90, I believe that the experiences of 3.90 flow into each of the following versions and so each newer version is "de facto" the most tested version.

That is no myth!!

That is a myth!!

3.90.3 --alt-preset extreme (blank media & hard drive space are cheap these days... so why not )

even if one version encodes a little slower(a few seconds) than others, what's the big deal

i see people saying stuff like "this version, that version is faster"... imo, they are plenty fast enough no matter what version you use

be patient and use the best quality version

Vietwoojagig: You should have included 3.96 in the poll, especially since there have been so many posts about this version recently (and quite a bit of of posted listening test results).
http://www.hydrogenaudio.org/forums/index....showtopic=19660
http://www.hydrogenaudio.org/forums/index....showtopic=19564
http://www.hydrogenaudio.org/forums/index....showtopic=19441
http://www.hydrogenaudio.org/forums/index....showtopic=19404
http://www.hydrogenaudio.org/forums/index....showtopic=19387
(and the list goes on)

I haven't done any MP3 encoding recently, and I don't really plan to in the near future, but if I had to encode right now I would probably pick 3.96, for the reasons guruboolez mentioned in this post.

Haven't you seen the 3.96 vs 3.90.3 regression thread? If things got worse, does it really matter which one has been tested more than the other?

EDIT: And really, if you test chunk of code, then rewrite the code, is the testing you have done still valid for the rewritten code?

Isn't 3.96 beta? Shouldn't that be considered "unsafe" to use?

But 3.96 is still beta. Only for testing.

As the person who got -Z included in aps I don't buy his reasoning. It does need to be used lots and tested lots to find any bugs. But that doesn't mean it should be used for "archiving" - and I think there are enough posts in the regression thread to suggest that this version needs further tweaking. Though there are very encouraging results too!

I'd say that if any version can be shown to improve more samples than it damages, and for the damage to be smaller than the improvement, then it should become the recommended version. EDIT: At the same or lower bitrate!

I'd suggest that pushing for a new recommended version without this is just silly.

Test and use: yes, to help development.
Recommend for quality archiving: no.

Cheers,
David.

3.90.3

LAME 3.90.3 with --alt-presets


quality is more important for me than encoding speed.

OK, life is short. Time is money. And the early bird gets the worm.

All that said, this is why I use 3.90.2 with -Z

1. I encode it once
2. CD-R media is cheap
3. I will listen to it for years if properly done.
4. I hate reencoding after I found out that version x.x.x.x had a bug that was missed. To date, each time I embraced a new version, it was found to have more problem samples than 3.90.2.

So I guess my question is, what is the hurry. So it takes a few more minutes to encode a track. A friend of mine is sad now because his MP3s that are old Xing sound like crap on his new speakers. Oh well, it sure encoded fast, and look, the files are smaller.

Actually, I use both -- (at -ape):

a) 3.95 Lame_enc.dll with EAC and/or CDEx for "everyday" encoding -- because speed and size DO count, esp. if you're offloading to an 128MB MP3 player.

b) 3.90.3 external encoder for serious archiving

... although I really can't ABX between them at -ape anyway. So I'm thinking towards the day when I have serious-enough equipment that it might make a diff... and by then my hearing will be gone.