LAME 3.90 vs. LAME 3.97

2007-02-27 12:48:17

Okay, I gather this is an easy one but I'm pretty clueless about it.

Are there any audiable differences between music tracks encoded with LAME 3.93.1 and those encoded with say LAME 3.97?

I'm really not sure what it means when they release these new encodrers.

Thank you...

LAME 3.90 vs. LAME 3.97

Reply #1 – 2007-02-27 12:57:43

Take a look on the changelog:
http://lame.cvs.sourceforge.net/*checkout*...ml/history.html

The audibility of the changes depends on the listener's ability. Personally, I can ABX the difference on many settings and samples.

LAME 3.90 vs. LAME 3.97

Reply #2 – 2007-02-27 14:12:46

I haven't sufficient experience with 3.93 to talk about but with 3.90 which is in your topic.

Until relatively recently I used 3.90.3 ABR very high bitrate cause with samples that are very problematic to 3.96 and 3.97 things were better using 3.90 ABR. But take it with a grain of salt: using 3.97 ABR very high bitrate makes them pretty acceptable too.

These samples belong to problem classes where 3.96/3.97 flaws in a special way and which are covered by current 3.98 development. If you look at serious pre-echo problem eig, 3.90 behaves worse to my ears than 3.97 (though it's bad with 3.97 too).

That's for high bitrate usage where to me things are a bit in favor of 3.90.
If you're out for efficient encodings using for instance -V5 or -V4, 3.97 may be your best choice as quality at this quality level is great for most users with most music.

My current mp3 favorite Helix is well worth a try. It's robust against artifacts (for instance eig performance is a dream compared to other encoders) and has a very good VBR method. Use for instance -V65 for the moderate bit rate range, or -V120 for the high bitrate range (~200 kbps on average). Drawback in the high bitrate range is the very high frequency behavior which is worse than Lame's - but it's rare that it is audible (may be different for young listeners or other people with very good HF listening abilities).

LAME 3.90 vs. LAME 3.97

Reply #3 – 2007-02-27 17:31:30

Thanks guys... 

guruboolez, I will check that link.

And halb27, thanks to you too, and yeah - I made a mistake in the title. Should have read LAME 3.93 vs. LAME 3.97.

I used that as an example. I'm using a CD ripper with LAME 3.93.1 (Dibrom's switches) and I wanted to know if 320kbps/joint stereo files ripped with that are "worse" (audio quality wise) than those encoded with a higher version of LAME. I have no way of checking it myself, and well even if I did I wanted to know what the experiences in general are.

And if you could recommend a good freeware ripper that uses the "best" LAME codec that would be great (I am using dBpowerAMP10 at the moment).

LAME 3.90 vs. LAME 3.97

Reply #4 – 2007-02-27 17:42:26

Have you looked at the FAQ?

All your questions are answered, and you'll find the general consensus, not just halb27's personal opinion.

Cheers,
David.

LAME 3.90 vs. LAME 3.97

Reply #5 – 2007-02-27 18:35:26

• I remember gameplaya using 3.93.1 at low bitrate and a discussion about his setting which was a little strange to many members including me. I was concerned about version difference towards 3.90 then and IIRC 3.93.1 is pretty much the same as 3.90.3, and using CBR320, there shouldn't be a remarkable difference. However with CBR320 there aren't essential differences anyway between the Lame versions.
• As a consequence I wouldn't change 3.93.1 against 3.97. If I would change at all (no real need to) I'd change to 3.98. It's on alpha11, but has improvements over 3.97 already.
• Be careful with the version history. You never know for a fix towards what it applies as usually there's no reference to it. If you take all the fixes after a certain version it does not mean that these fixes apply to a problem that existed in this particular version. IIRC there have been some problems introduced in version around 3.94, and you will read about fixes concerning this. Sure these fixes don't affect 3.93 or 3.90 behavior.
• It's up to you but using CBR320 is not wise to use for most members here. mp3 quality is so high with a lower quality setting that you don't really benefit from that. In the rare cases where quality suffers at a moderately lower setting (pre-echo samples) you will not get a really better quality using CBR320. Try eig to hear for yourself.
  If you use ABR (a less popular vbr method than VBR which according to experience is more robust) at high bitrate, for instance --alt-preset 270, you can expect the same quality in practice. 270 kbps is already very high and offers a big safety margin, and you can go quite a bit lower without really sacrificing quality.
• I used EAC as a free ripper before I switched to current great dBpoweramp. You can cofigure EAC to work with any CLI encoder and thus with any Lame version. Current dBpoweramp isn't free but is not very expensive IMO, and I absolutely love it's way of ripping. Secure and usually fast. If you can afford it I'd buy and use it. I personally don't rip and encode in one step but encode the wavs to ape and mp3 (or whatever I  like) using foobar. If necessary I also do some wav editing before (clueing together tracks from a live CD for instance, or do some fading in and out).