Lame 3.90.3 vs. 3.96.1 - 3.90.3 FASTER?



I know this has been a topic of discussion in the past, but I’m having some interesting behavior I’ve noticed with Lame encoding. Everyone swears that Lame 3.96.1 is much faster than 3.90.3, though 3.90.3 is the preferred encoder due to its more thorough testing. When I encode using the command line option in EAC:
( V1 -b192 -F -q0 -mj )
3.90.3 seems to encode much faster!!! In fact, play/CPU indicator in dos window is usually ~7x as opposed to about 1.6 with Lame 3.96.1. So what gives, is 3.96.1 just optimized for preset settings, but actually much slower when it comes to user input configurations? Anyone care to offer an opinion as to which Lame I should be using with the above settings if the only concern is quality, not speed? Thanks,

Incidentally, 3.96.1 yields larger files at the same settings. When I bump up V1 to V0 using 3.90.3, resultant files are still smaller than mp3’s from 3.96.1 setting V1 (and resultant avg. VBR slightly lower too, e.g. 209 kbps vs. 214 kbps). Should I be at all concerned that 3.90.3 is actually running 5x faster, or should I just go with it and enjoy my new found free time?


This is because q0 in 3.96.1 uses a completely different algorithm than q0 in 3.90.3. I’ve heard that q1 in 3.96.1 is approximately equal to q0 in 3.90.3. Your settings look pretty much like --alt preset standard. Why don’t you just use that? Or if you want higher bit rate use --alt preset extreme. Finally, why do you have a strictly enforced minimum bit rate of 192?


Elder Young: Thanks for the response. Actually, I just picked those settings because to me it was the best sound quality vs. disk space ratio, IMO. Obviously, people have different opinions on the matter. I didn’t want anything below 192, and I want to give lame the flexibility to go to 320 whenever it needs to. Strict enforcement to 192 because I’ve heard some players choke on really low bitrates, so for greatest compatability…I’m backing up my CD collection (~400), so I’m trying to do this right the first time. Let me know if you have any tips…


Well the main thing I see is that you’re your files are encoded at a very narrow bit range. See here:

Basically you’re encoding files with an overall bitrate of about 210kbps and a minimum of 192 kbps. If you want to have a fairly narrow range like that you could just use: –alt preset 210 which will give you an average bit rate of 210. You can change this number for a different average bit rate. The problem with this is it will probably not allow the bitrate to go up to 320kbps unless you set the number close to 320.

The other option you have is just to use –alt preset insane, which will record at a constant bitrate of 320kbps. It may be a waste of space, but you wouldn’t have to worry about low bitrates.

I suggest you encode a few of your favorite songs, preferably ones that are pretty complex and might require a lot of high bitrate frames, using the –alt preset standard, –alt preset extreme, and –alt preset insane to see if you can tell any difference. If you can’t tell a difference on complicated songs then you probably won’t be able to tell a difference on simpler songs. The settings you use really depend on the individual song, the types of music you listen to and how good your hearing is.

I think I have pretty good hearing and the few times I’ve compared files preset standard files to preset extreme ones I can’t really tell much difference. Most people use preset standard. I’ve encoded all mine at preset standard and the bitrate can vary a lot. For instance, Weezer-Butterfly is really simple acoustic song and only encoded at 134, whereas Weezer-Glorious Day is more complicated and encoded at 234.

Here’s an explanation of some of the presets:

If you really want to back up your CD’s you should encode using a lossless format, which MP3 is not. I’ve never used one, but I know Windows Media Player 9 and 10 have a lossless format. Unfortunately it only compresses files to about 50-60% of their original size, so a long CD will end up over 400MB. Here’s a forum you might want to check out as far as other lossless formats.

I hope some of this helps and doesn’t just confuse you more.


Hey thanks again. Actually, I’m ripping images of the CDs at the same time, so I’m not too concerned about lossy vs. lossless, as I will actually have both formats available. This is all pretty much in preperation for a megadump onto an IPOD, pardon the expression. I realize most of my MP3’s end up with a narrow VBR range, which is not a problem to me, as I said I just don’t want things dropping below 192, and if there is a particularly complex piece of music, let it run at 320. My question to you is, whats the big deal about the presets anyway. Aren’t they just scripts that eventually feed some specific set of commands to Lame anyhow? They don’t run faster than command line options, to my understanding. So if I can tweak the command line options to my liking, which are then fed into Lame via EAC without my participation, its not really anymore complicated than using presets :slight_smile:


I think the presets have code level tweaks that can’t be achieved with normal switches like -V1, etc. Apparently for version 4 of LAME the programmers are switching it so the V settings do the same thing as the presets. For example V2 (or maybe V3) would do the same thing as preset standard.


I guess the “code level tweaks” was the answer I was ultimately looking for, though the only place I actually saw this explained was in the Hydrogenaudio MP3 forum where the Lame developers hang out and ruthlessly bash anyone who wants to know why presets are better than command line options. I actually wouldn’t mind using “preset extreme” if I could specify a minimum bitrate. In any event thanks for all the help, if nothing else I learned alot about how some of the newer features of lame work.


You can specify options after the preset.

I think you’re wrong in wanting to set the minimum bitrate to 192kbps. 128kbps as minimum will give Lame more flexibility and will probably also be more compatible with hardware players.

So what you’re looking at now would be –alt-preset extreme -b 128 -F in Lame 3.90.3.


I thought the presets did force -b 128 as one of their options, to deal with occasions when the ATH and other parameter would allow a lower bitrate, but that would not be adequately covered by it.

Even with a minimum bitrate set, it can still drop below that on “digital silence”.


I don’t know if this will help, but I’ve attached an image of the verbose output from LAME using preset standard. I think some of the settings are exclusive to to the song I’ve encoded.


I wasn’t thinking when I forgot to post the preset standard image. Here’s it is.

Here’s the same song with preset extreme.

According to WinAMP the preset standard file is 229kbps and the preset extreme is 284kbps. This is a lot larger difference than I expected.


Don’t listen to this guy. Anyone still using LAME versions older than 3.96.1 is talking crap. The alt-presets are therefore pretty much useless. Based on retarded tests done by stubborn idiots, who, once we tried to throw in some clearly opposing test-results, simply denied us all access to the entire hydrogenaudio domain. That’s how guys like this NR-something operate;
Once you don’t agree, and you have the data to back up your statements, they try and tape your mouth shut. Especially if they suspect they indeed could be wrong. Fools.



Now there is a clearly formulated objective argument. No wonder you were banned at HA - those idiots simply could not take your inexorably logical line of reasoning. I hope you did save ABX logs of your tests, though?

Tiki, I can second Elder Young and NRen2k5’s advice. Specifying 192k as minimum bitrate does not make sense. But if you insist go for it, it should not hurt anything besides your hard drive space. Might as well go all the way to –alt-preset insane.


What purpose would that serve, exactly? They’re dedicated to getting the best possible results out of different audio software and hardware. They could care less who’s wrong and who’s right in the end.

It sound to me as if you’re sore because you’re proven wrong too often.