What do the VBR quality settings, such as V 0, do compared to the algorithm settings, such as q 0.

Also, I have been using V 0 and q 0 with LAME 3.93 and I was able to encode at about 5-6x "under the play/cpu header", but using the same settings in LAME 3.96 I can only encode at about 0.6x. Additionally, changing to q 1 results in the same encoding speed as I had received with version 3.93 at q 0. Is there really that much difference between q 0 and q 1? I apologize now if the question has already been answered, but I couldn't find it using the search topics.

It is recommended that you use the built-in LAME presets rather than creating your own command line. The recommended LAME settings to use are here.

The q settings determine how hard (and for how long) the encoder thinks about encoding something properly. So a lower q = longer encode times but better quality.

The V settings determine how many bits the encoder is going to throw at the situation. Lower V gives a higher bitrate for better quality.

-q0 in 3.96 added the new huffman divide stuff which drastically slows down encodes. All the old q levels were bumped up by one.

Your MP3s will sound perfect though with LAME --preset standard, what you're doing is both overkill and probably sounds worse.

I agree that it is probably overkill to use these settings but can someone elaborate why it would sound worse (other than a 3.90.3 vs. 3.96 issues)? Some of us have more cpu time than we know what to do with

The alt-preset was testet ,tweakted and tuned for over several months (years) .
It was testet by many people and with objektive listenings test's.
it contains code leve tweaking not able to do with home made command lines.


try reading the FAQ

Sorry for the confusion, I thought the question was overriding the setting in aps or ape (etc.) with -q 0 and -V 0. Looking back i realize that was never stated. So I'm not sure if I should spin this off into another thread, but is there a sound quality (not speed or efficiency) argument to use the basic 'aps' over '--alt-preset standard -q 0 -V 0'? Again I am hoping to stay away from the 3.90.3 vs. 3.96 comparisons. Thanks.

Sorry I didn't post my full command line options, but here they are for anyone interested
-b 192 -m s -V 0 -B 320 -q 0. I just used these settings and copied the files as compressed files, directly in EAC, using LAME.exe as an external encoder.

I encoded Portishead's "All Mine" once using the -alt preset extreme setting and another time using the above settings. I listened to them back-to-back with my Grado SR60's without knowing which version was which. The version using my settings had a very, very slight bit more detail, probably because the file ended up an average of 254kbps versus 215kbps for the preset, but the vocals were easier to distinguish with the preset. With the preset it sounded like the vocals were a separate layer on top of the beat, rather than all one level like it was using my settings. Overall, the vocals sounded more forward with the preset, which is something that some people like and other might not. I actually like it. Just thought I'd share some subjective results of my "testing."

Before someone comes and starts screaming, I'll recommend you take a look at TOS #8. Your test wasn't a true blind test. You need to test versus the original also. What if you're command line changed the sound, and it just happens that you like that?

Sorry dreamliner77, I wasn't trying to make any kind of overall judgment or scientific statement of fact. I was just trying to say that with the settings I used and only comparing the two versions, I had an opinion. I wasn't trying to bias any one else's opinion so I apologize for any statements I made without evidence or facts. I will go look over the rules so I don't make the same mistake again.

No apology needed, I was just hoping it may drive you to do an actual abx test.

The presets are there for a reason, they were EXTENSIVELY tested.

-v = amount of bit to use for higher quality
-q = amount of encoding time to use for higher quality

I've always used --preset extreme and have been happy with it. Still, there are some "problem" samples in my music which I ABXed with the original and the LAME encode, and I was able to tell the difference. It wasn't easy but I passed every test and decided that was enough for me after 8 tries.

Enough to tell me, yes, I can hear the difference, with my headphones and sound card, but no, I'm not going to encode with --preset insane because the difference isn't enough to bother me. I have tried other codecs that did far worse, to the point where the result actually annoyed me on those particular samples, and that is what started me comparing codecs and eventually ended up with LAME.

Until I learned a bit more and re-ripped my CDs to lossless FLAC. Then I hope to use that to generate and compare any of the lossy formats easily.

But I can understand the confusion. There have been several posts mentioning the -v settings which are equivalent to the --presets, and also posts mentioning that users should expect the authors to be standardizing on one set of commandline options vs the other. It really IS confusing and so I'm not surprised at all when someone comes on and says "Which options should I use" because if they have done any scanning through the forums, there is a lot of conflicting info in there. It's only natural to ask as a result.

My advice to the original poster: stick with the -presets. To decide which one to use, either find some "problem" samples in your own CDs, which is preferred because that's what's you'll be listening to, or, use some of the known problem samples found here on the board. Do some ABX testing... I used Foobar2000. Compare the original wav or lossless with the lossy version. Find out what your ears and listening equipment can really do, and, you may find out what you "need" is a lot less than what you thought. With some PX100 headphones and an SBLive sound card, I could tell the difference between a sample's wav and --preset extreme mp3. But it was so slight it was insignificant to my hearing enjoyment. Do I need --preset insane? Not at all. I might even get by on standard for most stuff, and should probably test it next just to see if the difference is enough to bother me or not. But disk space isn't an issue for me, so this is more of a curiosity thing than worrying about disk space for listening. I got a Nomad Zen Xtra 60GB and so far it can hold everything I have. If only I could convince Creative to support FLAC or OGG or other gapless, open formats, heh heh.

If you're looking for the closest thing to an exact copy of your CDs, I recommend looking into lossless formats. I like FLAC because so much software and hardware supports it, and it's open. It gives me a nice backup of my CDs, gapless playback on my PC, and lets me encode to just about any other format without worrying about transcoding errors, or pumping CDs in and out of the drives which is a pain. Do I need FLAC because anything less means I can't enjoy my music? Not at all. But it's those other conveniences, a backup, easy source for conversion, never having to tag again, etc, that make it really attractive to me. Wish I'd know better before I re-did all my MP3s.

Best of luck on your testing and ripping.

In 3.96

--alt-preset standard -q 0 -V 0

equals (--alt-preset standard == -V 2)

-V 2 -q 0 -V 0

equals (-V 2 -V 0 == -V 0)

-q 0 -V 0

equals (-V 0 == --preset extreme)

--preset extreme -q 0

which is fine to use.

In 3.90.x this equation is not true.

In 3.96

-b 192 -m s -V 0 -B 320 -q 0

equals

--preset extreme -q 0 -b 192 -m s

You should shorten this to

--preset extreme -q 0 -b 192

or

--preset extreme -q 0

or

--preset extreme

or

-V 0

Really? I understood the presets did not map directly to any set of command line arguments. That might have changed with 3.96 though.

Since 3.95 all presets are mapped to the V-switches

Watch this

please tell me if I'm wrong, but i think the old q levels were bumped by two, no? Lame 3.96 uses -q2 now.