So, I finally decided to give Lame 3.96.1 a try.
A few questions now.

First of all, is it safe to use? I mean, will the audio quality be worse than the 3.90.3? Is 3.90.3 still the best compile?

And second, after trying it, I discovered that --alt-presets now allow lowpass filter modifications (and turning those off as well). How come this occurred?
(Yes, here I go again with the frequency ranges)

Many thanks.

http://www.hydrogenaudio.org/forums/index....showtopic=28123

this should answer your question?

2. LAME 3.96 is ~ 2X faster than LAME 3.90

At high bitrates (apm & aps):

3.90.3 = slow, guaranteed safe
3.96.1 = fast, probably safe, slightly smaller filesizes

Not picking on you, but people around here need to let go of this long-held assumption. Neither 3.90.3 or 3.96.1 are "guaranteed safe" using aps and ps, respectively. Both can be ABXed on a number of samples quite easily. However, both should give you transparent results most of the time. They could be considered the best compromise between "transparency frequency" (how often they give transparent results) and file size. Personally, I use 3.96.1 -V 1.

Somehow 3.90.3 aps obtained an almost religious following around here, with Dibrom reluctantly receiving (from what I can tell) God-like status as its main tuner.

Very good point freakngoat.

3.90.3 is stil the recomended version, but even so, it is still MP3. It is not perfect. It is not transparent on every sample. I think that people need to be reminded that ALL lossy encoders fail on specific samples. Even 3.90.3.

Dibrom deserves a lot of credit. Still, there is only so much MP3 can do.

Since somehow mp3's somehow became mission critical files and we be dammed if we had an inferior compressed version of our music were so crazy about I gather the
quest and topic of 'Which version should I use' will never cease. I pick on nobody more
so than myself as luckily being a reader longer than a contributor watched everyone
else do the dirty work with countless requests to know we are all a little overly obsessed
with it .

I understand Dibrom is HA's founder and personally did extensive tuning on 3.90.3 giving him the right to post that. However besides probably having a life outside of mp3
and maybe only he could answer but does anyone know why he doesn't list it as the
recommended version?
I gather even so that may not slow down the countless requests and might add more
to why so... just thought it might help the people needing a push to finally 'progress'

Definitely fitting for this topic

Regards

there is a thread (which I can't find right now), where LAME 3.90.3 vs LAME 3.96 was tested. On some samples LAME 3.96 was better on others LAME 3.90.3 - however, LAME 3.96 was overall better, especially when using --ps

I use LAME 3.96.1 and love it

cheers

The more I think about this, the more I want to say something...
*Getting on soap box*

Not just MP3, but specifically alt-preset standard. Not only is it not guaranteed to give you transparent results, but the artifacts I hear on certain problem samples are too annoying in my opinion to trust it completely anymore. There was a lot of pressure to get the bit rate down when the preset came out from folks stuck on the r3mix setting. After my own testing the higher quality 3.96 -V 1 setting replaced 3.90.3 aps as my setting of choice. While I can still ABX it, the artifacts are quiet and unobtrusive on the most difficult of problem samples (see this thread)

Anyway, that is completely my own opinion. People can use whatever they want, and 99% of people will be totally happy with (a)ps. I guess the moral of the story is to do your own testing and find out what works for you, and to be careful with words like, "guaranteed."

Careful, I'm not sure that was the conclusion which was reached, if you're referring to the thread linked to in my last post.

The general conclusion I think we reached is that they are about equal in quality, give or take. But as you can see, there wasn't a whole lot of testing, barely anyone participated. This has a lot to do with the fact that ABXing at this high quality level is difficult. It really is hard. But that's a good thing because the idea is to get transparent results at this level--you're not suppose to be able to ABX it. The conclusion I reached is that neither are good enough.

well, in terms of --ps LAME 3.96 won 7 times, 3.90.3 only 5 times...1 time it was equal. Besides that, even if it they produce a equal level, LAME 3.96.1 is still faster, fixes a bunch of bugs and produces most of the time lower bitrates...