Now a lot of users already have multicore systems. And LAME is enough fast on those systems. Is there any chance to get LAME version with higher quality at cost of encoding speed.
I will use it if there will be a few percent better compression ratio at cost of 2x (maybe more) down speed.

A switch -h doesn't do any real deal.

Probably it has already been discussed but I haven't found anything.

First thing comes to mind is use of the old VBR algorithm. Correct me if I'm wrong, vbr-new was develped to be faster, but not necessarily better quality.

False. The new VBR-Algo was adopted as the "recommended" one not because of speed advantage, but because in widespread DBTs it performed better qualitywise than the old one. Thus, the old VBR-algo has worse quality AND speed.

A more relevant question to the original poster is:
Why are you asking for higher quality? Do you notice any problems? If you dont notice any problems, then how can quality improve?

P.S.: Another interesting question is - why is mp3-encoding so fast? Perhaps because compared to other modern codecs, it is relatively simple and well understood? Are computers fast at doing simple things? Are well understood things easier to optimize?

Actually, my memory from lurking older discussion threads was that vbr-new is faster and of equal or better quality than the old algorithm.

Isn't there a 'q' quality setting that does something?

To my knowledge, vbr-new was developed to be faster than vbr-old while providing the same quality. I think the file size of vbr-new is a tad larger as well. I don't really remember though as the last time I did file size tests of vbr-old vs vbr-new was back when 3.97 was in b1 phase. Still, vbr-new should provide the same sound quality as vbr-old.

The -q values no longer really apply in Lame 3.97 and above either. Correct me if I am wrong but with Lame 3.97, -q0 should give you the same quality as -q3. I think that the -q switches are pretty much completely disregarded with Lame 3.98 (which is still in beta) when using -V 4/5 or higher.

I don't think there is much you can do to reduce encoding speed and increase quality other than increasing the VBR setting. Still, you should conduct some ABX tests if you suspect that something is wrong with your current Lame setting. Otherwise any flaws that you hear fall ill to the placebo affect.

----------------

No, I don't complain about quality. I just don't mind to spend more time to encode via foobar (dual session) on my dual core. Trying to reduce bitrate to put more albums to mp3 player.
My cds (loud rock and metal) are encoded to V5 new at 140-150 kbit/s. A quality is ok for me but the bitrate is still high. Something like 130-135 kbit/s with same quality will be welcomed.
V6 isn't acceptable. It easy to abx.



He he, nice try. I already keep my rips in lossless. But I still need to fit more songs in my limited space mp3 player.

-----

Yes, it does.
But I need particularly LAME. And I have a reason for that.
I'm 3 years member of HA. I already know enough about Vorbis, AAC, lossless, etc.
Please, can we focused on LAME now?

--------------

Please provide information for your described sudden "quality-jump" from V3 to V2.

EDIT: V3 actually may be an interesting alternative for the original poster, if he previously used V2. The main difference between v2 and v3 is that beginning at V2, the encoder spends significantly more bits to encode ultra-high freqs - stuff at ranges, which most people dont notice in normal music - this is why there is a bitrate jump between V2 and V3. So, unless you have bat ears, you can significantly reduce bitrate by going from V2 to V3, without significant degrade in transparency.

--------------------

I will ask admins to keep MY topic clean. No other lossy neither lossless codec talk, no flame wars about V2 and whatever stuff.

The name of topic is "Slower but better version/settings of LAME?"

-------------------------

I can clean this up, but you do realize you're basically asking the Lame developers to create a brand new psychoacoustic model?

This thread would be kinda pointless if I didn't allow Bourne and Lyx duke it out over V2 vs. V3.

EDIT: Added quotation.

In other words, you have no proof for the claim which you made? (which is that at V2, lame is okay, but below it (i.e. V3) there are way more samples with problems). May i remind you about TOS#8?


The reason why i mention V3 is because in this thread you so far mentioned two goals:
1. better quality at same bitrate
2. lower bitrate at same quality

There is no solution available for "1.". V3 however may perceptually be the answer to "2.".

HE started the topic. Maybe YOU should be the one to get outta here.



That would be great if he were using V2 but couldn't abx V3, but...

-------------------------

Cut the guy some slack. Lossless codecs offer users the choice of longer encoding times for better compression; it's not much of a stretch to ask if there's a similar option for lossy codecs such as LAME. And as a matter of fact, there is:

Now, I don't know if that option produces lower bitrates or how effective it is (the man page suggests it's overkill), but the OP's question certainly wasn't stupid.

When I talk about better compression ratio I didn't mean in lossless way.
It is possible to improve quality of sound perception by more sophisticated (slower) algorithms . In lossy theory (audio or video). Any lossy codec mp3, aac, xvid, h.264, etc.
But there is certain limit for all cases.

The question is:
Is the LAME already optimized maximally for its speed/quality balance and there can't be quality improvements ONLY at cost of speed? I'm talking about cost of speed only. Not quality improvements that are purely coming from absolutely new psy.

Try encoding with the following parameters using Lame 3.97 (you know, the recommended version) and compare the results:
-V5 --vbr-new
-V5 --vbr-new -q0

...and tell us what you find.