I want to rip CD's to .mp3 format. The .mp3's need to be 'top' quality regardless of file size. Which version of Lame is advised and what command line entry needs to be input into EAC V0.95 beta 4 to achieve the desired effect?

I'm a noob...be gentle with me!!

I'll be gentle although you could have done a tiny bit more research, have you seen this?

List of recommended LAME settings:
http://www.hydrogenaudio.org/forums/index....showtopic=28124

Advised Lame version is 3.97b2 at the moment although a switch to 3.97b3 may be imminent.

I would recommend to use latest lame (3.98a6) with -b 320 switch. This is the maximum quality mp3 can offer.

J.M.

Thanks for that. Please confirm that it's a simple case of exchanging the bold section of

-V 0 --vbr-new --add-id3v2 --pad-id3v2 --ta "%a" --tt "%t" --tl "%g" --ty "%y" --tn "%n" %s %d

for

-b 320

thanks in anticipation of your patience and further help.

Actually the exchangable part is:

-V 0 --vbr-new --add-id3v2 --pad-id3v2 --ta "%a" --tt "%t" --tl "%g" --ty "%y" --tn "%n" %s %d

-b 320 --add-id3v2 --pad-id3v2 --ta "%a" --tt "%t" --tl "%g" --ty "%y" --tn "%n" %s %d


-V 0 --vbr-new is the highest quality VBR preset.
-b 320 is 320kbps CBR.

These two settings should provide very high quality but -V 0 --vbr-new will produce smaller files.

Your a gentleman! Very clear. Methinks that the previous two posters assumed too much regarding the level of my knowledge and understanding!! You're not a teacher are you??!!

Yeah, let's throw all common sense out the window and let's use CBR0rz 320kbps, shall we! Weeee!!!

Tbh, there seems to be an attempt by many techies to over sophisticate the matter. Which setting is best? Well, it seems to me, having read many threads both here and elsewhere, that above a relatively modest level of resolution, the majority of us can't hear the difference between the options.

It reminds me of the popularity of vinyl records in the last century, when the techies then would develop better and better players and peripherals until the wow and flutter values became academic...coz you could hear the difference. But, it didn't stop them from continuing to imply better and better audible quality as new kit was developed!!

If my supposition is correct, then it is left simply as a matter of informed choice, largely based on which bits of what you heard or read convinced you most.

Presumably, if there is an obvious choice for all of us (for best quality results), then I would expect it to be presented as this forum's adopted official guideline.

Yup, -b 320 is prolly overkill, anyway I'd add "--ignore-tag-errors".
Actually there is an official guideline: you need to achieve transparency, and that is relative to your ears, anyway most of the times -b 320 is a waste of space, matter of fact it's also called "--preset insane".
Now go ABXing mad, pal ;] !

If you were a betting man, which command line setting would you choose as a world standard, to be reasonably sure that even the most discerning audiophile would not be disappointed?

Post back with the code and I'll input it forthwith!! Simple cut and paste job, yer know what I mean?

FLAC.

But if your only option is MP3, i think that -V2 -VBR-NEW is a good bet.

Here's me bet:

-V 0 --vbr-new --add-id3v2 --pad-id3v2 --ignore-tag-errors --ta "%a" --tt
"%t" --tl "%g" --ty "%y" --tn "%n" --tg "%m" --tc "LAME 3.97b3" %s %d

jmartis' suggestion for 3.98a6 -b320 is fine if you don't care about file size and want to make sure best possible quality. To get an impression of what can happen to natural music with settings like -V2 you may want to try trumpet, herding_calls, harp40_1 from the problem sample thread. Even -V0 is a bit problematic, especially with Lame versions prior to 3.98a3.

CBR 320 sure is overkill most of the time, but it gives a very high degree of security. An altenative without losing too much safety IMO is to go a bit lower and use for instance CBR 256 or ABR x with x ~ 256, that is to use a parameter like --alt-preset 256 or similar. See for instance my signature with the Lame 3.90.3 setting I use.

Quite a lot of members here on HA have a paradigm like 'VBR is best (in any case)' or 'very high bitrate is insane'. This is valid opinion, no universal wisdom. And it depends on how people feel towards encoding results like that of trumpet etc. Not everybody does care about samples like that.

It's now been installed. I'm trusting in you!

LOL!
Yeah it would be nice to see some typical of HA.org recommendations such as using
-V 2 --vbr-new when "most folks wouldn't be able to tell the difference" between that and a wasteful 320kbps mp3.

You now understand the problem for the newbie. Guys (regulars on here) present with an authoritative stance and prescribe the best solution to a problem. The new problem becomes choosing between the different recommendations.

Who's right, is it kwanbis, DARCode or halb27?? Three different answers to the same question. Methinks that this subject is like politics. No 'right' answers just informed opinion.

I'm now waiting for the next alternative reccie!!

because if you have to be 100% sure, then, FLAC/WavPack (lossless) is the only way to go.

If you want to be "reasonably sure that even the most discerning audiophile would not be disappointed", i would go with -V2, is like a compromise. I can't tell V5 apart from the original.

Also, audiophile is not the right word, i consider myself a person that loves audio, but i don't have "golden ears". I think "golden ears" is a better term.

I too am relatively new to this stuff and originally came to this forum in search of the "best" way to create MP3's. I discovered, though, that for me it made more sense to rip to a lossless format such as flac, because you then have the option to transcode to any other format if you wish. So, you only have to go through the ripping process once. That way if your needs or preferences change in the future (regarding compression format/configuration) it's no big deal to redo your entire collection simply by a batch transcode procedure ... rather than re-ripping everything.

dogbreath, the only way you can know which is best, would be by running ABX tests yourself. Without training, it is almost garanteed you would never find a problem at -V2 --vbr-new. Even with training, you wouldn't on 99% of samples. Most likely, by running ABX tests, you'll pick a -V level well below 2.

-V0 --vbr-new, and -B 320 are, generally speaking, overkill. Unless you don't care about bitrate. But then why are you using a lossy codec at all?

In conclusion, use -V2 --vbr-new if you're lazy, use -VX --vbr-new (where you choose your own 'X') if you want to run some ABXes yourself, or -B 320 if you're extremely anal and don't care about hard disk space.

Seems that my last 128 kbps test was useless when reading the recommendations here...

Thanks for taking the time to post. I'm going to give -V0 --vbr-new a trial. Disc space isn't an issue. But I accept that there's little point in encoding silence at 320kbps. I don't need to live on the edge of acceptability, so I'll err on the side of caution and indulge in a bit of overkill.

You ask why use a lossy format at all. I have a Creative Zen but, more importantly, an in car sound system which supports the .mp3 codec. They're the prime reasons. In fact, regardless of the format in which I acquire music files (.mpc, .ape, .ogg, .wav, .flac), I always convert them to .mp3.

It's like garlic bread. I've tasted it! It's the future you know!!!

Any test adds to validated experience on codecs. So your tests are very welcome. But any test only yields part of the story. And weighting usual quality against robust quality is the everlasting story people have different opinion about. Not everybody loves -V5 robustness. Ever listened to trumpet at -V5?

No, but -b 320 is overkill, so is -V0, -V1 and most of the time -V2. I think that using something between (and including) -V5 and -V3 is best.

I'd suggest --preset standard as the "standard". Though I think now its fashionable to call it --V 2. Same thing though.

It's not a year ago when I thought exactly like this. But I really invite you to encode harp40_1, trumpet and herding_calls with -V3 just for a very special experience. You won't beleive it if you don't listen to it. And it's pretty simple natural music.
Nonetheless everybody has to find his own special compromise. But I think it's good not to close the eyes with respect to this extremely audible failure of mp3 at moderate bitrate.

I'm with Sebastian here. The -V switches and the "low end" listening tests are there for a reason. Show some respect for the work and use them... or maybe trust is a better word.

But if you only trust 320 kbps CBR... then why not go lossless right away?

Personally I think -b 320 is a waste of space compared to FLAC or WavPack or any other lossless codec.

So, we're all agreed then!!

Nonsense. Your test and all others have provided information worth the read. Freedom of choice is good too. Education takes time. Thank you SebastianMares for what you have done here.

cheers,
herefornow

FYI, lame 3.97 beta 3 was solely released to better the halb27 samples...

So in case you wonder (and care), dogbreath, 3.97 beta 3 has now replaced beta 2 as the recommended lame compile since the fix for those samples was the only change done to the new version:


The "recommended lame compile" wiki page has a problem so here's the link to the rarewares page: http://www.rarewares.org/mp3.html

Those are problem samples. You have still failed to provide a single example from you music collection that would cause similar problems. So please stop making conclusion on samples that have zero real world significance. And even with those samples. I highly doubt you would spot the difference in everyday listening conditions.

And yes, I am pissed.

@Gambit
I totally agree.

Evaluation with only problematic samples result in biased conclusion. I wonder why he/she who did listening test don't realize this basic assumption.

Yup, even with the problem samples, it's not a foregone conclusion that they will result in "problems." I took a listen to herding_calls after it had created some hubbub in a previous thread, and I couldn't differentiate between the source WAV and a V2 --vbr-new encoding at all.

But I'm not about to start training myself to hear problems.

Ignorance is bliss... once you train yourself to hear problems you will start hearing them *everywhere*... it's like opening Pandora's box.

So, if you really want to enjoy lossily encoded files... do an ABX... but do not train yourself for the artifacts

Sorry, this is a bit too much. If they had zero real world significance why would LAME developers even pay attention to it? They do have a significance, but for most people halb27 seems to absurdly exagerating this significance. And what's worse for those people: each time some one asks for some uLRT4H4X0R2 settings for MP3 quality, halb27 comes around completely ignoring that most of these people are just as good of with something like -V4 or -V5. I agree that 3 problem samples probably isn't enough to completely base ones encoding habbits on.

Thanks again for taking the time to get involved.

I absolutely do care, otherwise I wouldn't be here.

I am merely perplexed as to why a group of seemingly very intelligent guys, even after many thousands of years of our species' evolution, can still quote opinion as fact and keep a straight face. If I have been remotely flippant, it merely reflects this observation.

I have drawn two useful conclusions from this exchange, the first being that there is no definitive answer to my question. In my naivety, I somehow hoped that there would be and it seems that many members of this forum erroneously still think that there is.

Secondly, by considering all the responses so far, I have chosen, with a fair degree of confidence, a ripping standard that will very likely exceed my expectations and requirements.

Many thanks again to all those who contributed. This is a really great place for a debate!

Your basic question of "what is the best mp3 setting " is flawed to begin with without further information from you about your intentions, i.e. How you listen to music (portable, slimserver, computer....) There was an answer given that lossless is the only solution that is proven to have no sound quality problems under any circumstances. The mp3 format will always have a handful problems samples that are not transparent. Some of us put more importance on those than others. The question is how many will you find in your collection of music. My guess is: not many.



So what did you choose? I use AAC because I mainly listen with an ipod, but when I used mp3 I ended up with V5 --vbr-new because I couldn't pass any samples in Sebastian's test with that setting.

I'd say those samples have almost zero real world significance and the LAME developers do care because a small chance exists that this problem has an impact in Music XYZ, Album ABC or Song 123.



I don't think so, I've listened to those samples (with 3.97b2) and it was (is?) quite a big issue IMHO. I couldn't ABX them at -V0 but at -V2 and I haven't even bothered to go lower for I knew it would get worse. But I strongly believe he should keep his samples to the MP3 Tech board, and not come up with it when people ask simple questions like dogbreath. Finding and ABXing problem samples is done for improving the encoder and not to base one's settings on them, that's what I think is right and believe many will agree. LAME has matured enough that telling newbies to tweak the settings isn't necessary.



I fully agree. And in case you encode not solely for your own listening using -V0 to -V2 won't hurt.

Thanks for taking the time to really listen to what I'm talking about. Not every opinion contributor here does so.
Sure - and I never said something else - it's up to everybody what conclusions to take. I do know my personal conclusions do not agree with the attitide of many members here. But in dogbreath's original post he made quite clear that disc space isn't an issue to him but top quality is, and in a later post he made clear that mp3 is vital to him. That's why I supported jmartis' 3.98a6 cbr320 suggestion resp. variants of it using a little lower bitrate.
I will not suggest this kind of solution of struggling for robust quality to newbies who don't make clear that they have a very high quality demand and don't care about file size.

Do you have to use the --strictly-enforce-ISO switch with 3.98a6 and -b 320?

if I'm not mistaken, this switch is on and forced by default from v3.98.4 and up

This is what I remember too.

-V 0 --vbr-new

I'm still waiting for that sample from your music collection...

I don't know why this important to you, but if you do like to hear it: there's nothing essential to mention.
I once searched through my music collection trying to find female voice samples that are hard to encode.
I found some samples which weren't encoded well at 128 kbps, but this was with a rather inferior setting which I choose in order to find such problems at all. When I used a usual setting at something like 160 kbps the problems were gone. Not real problems.
I had a hard time with a phenomenon about two weeks ago (and I still have it). One of the major reasons why I am about robust quality is that I don't want to keep a lossless archive in parallel. Taking into account my restricted hearing abilities and the quality achievable with lossy codecs, the appropriate way to encode to me is using wavPack lossy @352kbps (for very good tracks judging from a musical as well as sound quality background), or Lame abr @270kbps in all the other cases. Two weeks ago I encoded a Joe Cocker album. In this case it was a bit hard for me to decide upon which codec to use. I ended up with mp3, but decided to do an abx test on one of the tracks. I took 'Unchain My Heart'. I was pretty surprised that exactly these words in the beginning were't exactly like in the original. A rather subtle difference, but I got a 9/10 result.
But the strange thing is: I redid the test the day after, with no significant result. 9/10 is too significant to me in order to have it ignored, but if I can't reproduce the result it's worthless.

Anyway, if this all tells you I don't have the golden ears you are absolutely right.
But this doesn't bring me away from demanding robust quality. If I know a sample (harp40_1) that needs an mp3 encoder setting that yields an average bitrate of more than 200 kbps, I will use more than 200 kbps, and provide moreover for a good security margin. And when I see that Lame (before 3.98a3) and FhG encoders have problems with VBR on my favourite samples I will not use VBR, especially as VBR can hardly be expected to be advantegeous in the very high bitrate range I use. Maybe things become different with further progress of the 3.98 branch, but at the moment to me it's most appropriate to use 3.90.3 abr 270.

Sure my solution is not necessarily the only solution, not even to me. For instance there isn't a real reason why I don't use HELIX with level's setting requiring only about 204 kbps on average. Also there is no real reason for my Lame lowpass usage. It just doesn't hurt me as I can't hear such high frequencies, and it may do some good in problematic cases for the same reason why it's used with -V2, but I have no idea whether it really does. It's not essential, I just do it.

Sure I admire people like Wombat who obviously are able to find problems in encodings of their own musical collection.