Sounds sensible. I agree.
Sorry, my fault. I read the original text wrongly.

I would suggest "Audible differences between these presets exist, but are extremely marginal.".


I'm not sure where this is all going. I see that 3.97b is now official, so I'm not sure that the thread is going to change anyway...

Edit: updated my original post to reflect these suggestions.

Edit 2: I feel as though I'm jumping on the bandwagon a bit, by stealing someone else's idea and saying "Look at me, I've made it better!", but this is not my intention. I would simply ask that the thread, including the Quick Start passage, have clear, concise text with consistent formatting. On the whole, I liked benc/dev0's formatting for the VBR/ABR/CBR explanations.

I also really like the idea of utilising links on keywords - either pointing to a wiki page or thread on HA, or a wiki page on wikipedia.org. My suggestion would be the HA wiki.

So I have incorporated latest ideas.

The idea of linking keywords with wiki content or topics of HA is a good one, I just need hints here, which keyword together with link.
The wiki needs still a major workup, eg. see the eac ripping section about lame/mp3, where are the alt-presets dominant, still.

edit:

hm, btw. the recommendation to use MP3 for (S)VCD, low bitrates, is that possible ?
IIRC, (S)VCD requires MP2 (eg. toolame) by definition, standard ?!
hm, I think, if I am correct, that mp3 is not possible for S/VCD, then I think the mistake crawled in, as Kwanbis wrote about CD/DVD rip, and somebody else rewrote it to SVCD, never mind.

Thank you, though I have to say I still find the formatting confusing.

I only looked at Guruboolez's text. That used "1 CD DVD rip". I took "1 CD DVD" to mean a VCD or SVCD, i.e.: a DVD that had been ripped to a CD. It does seem that there is some confusion, certainly on my part! Maybe it should simply be removed. I think "Voice, radio, mono encoding, etc.".


Quick Start

Low Bitrate : voice, radio or mono encoding, etc.

For very low bitrates, up to 100kbps, ABR is most often the best solution.
Use --abr xx (e.g. --abr 80).

Portable : background noise and low bitrate requirement

-V6 (~115 kbps), -V5 (~130 kbps) or -V4 (~160 kbps) are recommended for this use.
-V6 produces an acceptable quality, while -V4 should be close to perceptual transparency.

High Quality : home or quiet listening

-V3 (~175 kbps), -V2 (~190 kbps), -V1 (~210 kbps) or -V0 (~230 kbps) are recommended.
These settings will produce transparent encoding. Audible differences between these presets exist, but are extremely marginal.

Best Quality : archiving

--cbr 320. This is the strongest setting for LAME MP3, with the lowest risk of artifacts.
With the exception of a few situations, quality is rarely better than the highest VBR profiles described above. There is a reason that --cbr 320 is called "insane". It is not recommended.


Please note - the link to "Artifacts" currently points to a non-existent page. Perhaps someone who knows what they are talking about could populate one?


Quick Start

Best Quality : archiving

--cbr 320. This is the strongest setting for LAME MP3, with the lowest risk of artifacts.
With the exception of a few situations, quality is rarely better than the highest VBR profiles described below.

High Quality : home or quiet listening

-V3 --vbr-new (~175 kbps), -V2 --vbr-new (~190 kbps), -V1 --vbr-new (~210 kbps) or -V0 --vbr-new (~230 kbps) are recommended.
These settings will produce transparent encoding.
Audible differences between these presets exist, but are extremely marginal.

Portable : background noise and low bitrate requirement

-V6 --vbr-new (~115 kbps), -V5 --vbr-new (~130 kbps) or -V4 --vbr-new (~160 kbps) are recommended for this use.
-V6 --vbr-new produces an acceptable quality, while -V4--vbr-new should be close to perceptual transparency.

Very Low Bitrate : voice, radio or mono encoding

For very low bitrates, up to 100kbps, ABR is most often the best solution.
Use --abr xx (e.g. --abr 80).

I meant DVD -> single CD.
For a 2-CD DVD rip, low audio bitrate is maybe not really ideal.

This is somewhat confusing as I've been using -V 2 --vbr-new for transcoding from FLAC. Are there any specific samples that are known to cause problems? I've never experienced the "artefacting" mentioned here

Wouldn't it make more sense to replace 'low bitrate requirement' with 'small size requirement'.
I think it's not the bitrate anyone cares about actually but the implications a certain bitrate has for quality and size. In this case (portables) it's size that matters, low bitrate is only a method to reach the actual goal: small size.

Makes sense, yes. I agree. Needs to be better worded I think ("size" is a relative thing - at least that's what I tell the wife). Something like "minimal file size requirement" - maybe "optimum file size"... maybe "minimal KB requirement"...

How about "low storage space requirement" ?

Regards,
Madrigal

I think these hints from magic75 should be taken anyhow to the explanations of the use of the new lame-codec switches. I find them excellent for people, who don´t know much about all the discussions about the best settings.

I'd say "minimal file size requirement" and "low storage space requirement" are the best options.

Let's bear in mind that most people actually reading this manual (and especially the Quick Start!) are not your average HA-readers or Lame-experts. They're bound to be people who've been pointed in the general direction of Lame by a friend, Google or some website because they we're told that Lame would create 'better' MP3's. That is why I think using simple to understand, straight-to-the-point, not too technical manuals is very important.

Users who want to create 22.500 samplerate, mono, VBR-new, lowest bitrate 173Kbps, highest bitrate 201Kbps, without replaygain, with lowpass 16000 and what-have-you MP3's will read on past the Quick Start for the more specialist switches.

What's up with 3.97b1?? Are we finally there where Xing was years ago???

I always used --alt-preset standard or --alt-preset extreme in 3.90.3 resp. 3.96.1 and besides small differences in transparency the sound has been excellent. And the frequency range up to the highpass filter entrance was perfect!

Today I've tried the 3.97b1 with --alt-preset fast extreme and -V 0

Both settings cut off all frequencies > 16 kHz most of the time - only above a certain dynamics level (~ -70 dB) those frequencies >16 kHz are saved during compression (tested all output files with CoolEdit)

What's going wrong here??? Is this because qval=3 is used instead of qval=2 (which was used in 3.90.3s preset extreme)?

How about you drop your graphic waveform software, take your ears out of the box from which they never served, and perform an ABX test? After that, report the results on a new, unagressive thread, and, if they're revealing, help LAME developpement through your experience!

I think _THAT_ would be a good idea.

Out of interest, could someone please point me in the direction of the test(s) which resulted in --vbr-new being adopted as recommended? Thanks.

• My tests
• Bug80

There are maybe other tests, but I don't remember them.

Thank you, guruboolez!

I knew there'd be one by you.

I've always used [--alt-preset standard -Y] to save space since I'm unable to hear anything above 16 KHz anyway. Now when I'm switching to 3.97, should I use [-V 2 -Y] or [-V 3]? I thought these two were identical, but they give me different filesizes.

Woah. I think you BOTH need to calm down.

WMM asked a fairly legitimate question, even if his tone was wrong.

I can't hear any difference, but it would be interesting to know why the lowpass is now more aggressive. This could make the sound degrade for the younger listeners amongst us.

This is the effect of the ATH. High freqs that are below the ATH are removed, and in the high freqs area, the ATH is very high.
Basically Lame is filtering freqs you can not hear.

-V2 is equivalent to --alt-preset standard, so I suggest you try -V2 -Y. I'm using -V2 --vbr-new -Y myself, my files come around 180kbps.

LAME sure has a lot of switches..... Just an observation. No opinions implied.

Relating to --vbr-new i have my old samples i already mentioned degraded with V2 regarding 3.96 vs. 3.97b1.
Using 3.97b1 -V2 --vbr-new with the birds and the sophia2 sample now should be easy "audible to really everyone" what problem i always talk about.
You may try this?

why was razorlame replaced with all2lame?

I'm not sure, but razorlame was choking on Replaygain messages from newer versions of LAME. Maybe that was why.

edit: grammar

Thanx for the post Synthetic Soul - post #70: How to get around lame halting if you pass an incorrect genre. It always seemed to get me when I was in a hurry or showing someone how great tagging was with EAC. I had to create another profile in EAC when I was compressing wav files without the CD. Now that problem is solved. Also I think the graph you presented in post #117 is awesome. A picture is worth a thousand words.

This post is great! Its up to seven pages now, and even though it is getting rather long, figuring out the settings to use with 3.97b1 is much shorter than just a year ago trying to figure out what the "ideal" settings were for 3.90.3 and 3.96.1. I'm swithcing from 3.96.1 to 3.97b1 and its only taken me this post to figure out what settings I am going to use, compared to the days of research it took in the 3.93 and 3.95 days. Most of the time this time was spent on considering to use --vbr-new or not.

I like the quick start guide concept, but I see a lot of discussion about the "recommended" settings for new users and what the aptitude of the reader is. I personally use -V1, but instead of defending that I would rather see the quick start guide simply recommend and encourage users to encode a song they are familiar with using several settings from -V5 to -V0 (even give the exact command lines to cut and paste) and playing them on whatever device is going to be used and encourage them to decide on a setting for themselves instead of trying to convince someone which setting they should use (this naturally encourages the defense mechaninsm). I think this will prevent a lot of back and forth of, "I was thinking of using this setting, is this a good idea..." and "what about this..." If you give them the settings to cut and paste they can just figure it out themselves. That's why I come to the forum; to learn what my options are, try them out and choose what works best for me.

I also think having an easy tutorial of what the command line options are is important and not too complicated for an HA reader. Because of this forum I use the following command line:

-V 1 --vbr-new --add-id3v2 --pad-id3v2 --ignore-tag-errors --ta "%a" --tt "%t" --tl "%g" --ty "%y" --tn "%n" --tg "%m" --tc "EAC lame 3.97b1 -V 1 vbr 210" %s %d

Thanx everyone for your input so I could come up with this.

Last suggestion; I don't see any mention of encoding audio books/speeches/mono sources. I tried several versions in the 3.93 days of vbr, abr and cbr with different settings, sampling frequencies, low pass filters etc. and could never match --preset-voice. It sounds awesome and gives about 25x compression. I'd like to see this setting in the guide.

I think it is normal in the learning curve of using EAC and lame with the power of the command line to question the "presets". I know I did, and when I participated in the forums I got slammed whenever I asked a question or presented something outside of the presets. I'm glad this forum presents the tagging commands, as it gives the new people something to chew on. So maybe an intro into how the presets were developed and how long they have been tweaked (years) might help squelch some of the "innocent" questions that come from people who obviously have not read much of the HA forums.

Great work people!

This thread is pretty long and I just got in on it, so I hope the following question is not a repeat.

Can I use the recommended settings for LAME 3.97 that's stickied in this forum for my LAME 3.96.1? I don't want a beta encoder, that's why I'm still staying with the older but "unbuggy" 3.96.1. Ideally, I'd like to use ~192 kbps for VBR (I used to use CBR until I found out that it's not as good as VBR). Maybe -V2 or -V3. Thanks in advance.

OK, I just ran lame from the command line for the first time. Previously, I used to encode my music as CBR via the LAME plug-in that I have for GoldWave. Anyway, here's the output. Can someone help me interpret what LR and MS mean? Thanks.

C:\Program Files\LAME>lame -V 2 --vbr-new "<directory_path\file_name.wav>" "<directory_path\file_name.mp3>"
LAME version 3.96.1 (http://lame.sourceforge.net/)
CPU features: MMX (ASM used), 3DNow! (ASM used), SSE, SSE2
Using polyphase lowpass filter, transition band: 18671 Hz - 19205 Hz
Encoding <directory_path\file_name.wav>
to <directory_path\file_name.mp3>
Encoding as 44.1 kHz VBR(q=2) j-stereo MPEG-1 Layer III (ca. 7.3x) qval=3
Frame | CPU time/estim | REAL time/estim | play/CPU | ETA
167050/167052(100%)| 4:42/ 4:42| 4:42/ 4:42| 15.447x| 0:00
32 [ 31] *
128 [ 2564] %***
160 [ 16901] %%*******************
192 [ 43138] %%%%%%%***********************************************
224 [ 53440] %%%%%%%%%%%%%%%***************************************************
256 [ 35980] %%%%%%%%%%%%%%%%%%***************************
320 [ 14999] %%%%%%%%%%%%*******
average: 223.3 kbps LR: 42165 (25.24%) MS: 124888 (74.76%)

Writing LAME Tag...done
ReplayGain: -7.7dB

C:\Program Files\LAME>

You can use anything you want, but its not recommended.



Left-Right stereo and Mid-Side stereo. Two different ways of representing stereo sound, sometimes one is more effiecient than the other.

suggestion: I don't see any mention of encoding audio books/speeches/mono sources. I tried several versions in the 3.93 days of vbr, abr and cbr with different settings, sampling frequencies, low pass filters etc. and could never match --preset-voice. It sounds awesome and gives about 25x compression. I'd like to see this setting in the guide.

Is there such a preset working in 3.97b1 ?
Does it deliver the best for the goal/bitrate ?
of course, then it belongs to the recommended settings.
Can you give the necessary informations regarding averaged bitrates, if it is mono or not ?

--preset voice is only available in the command line front-end, and is there for compatibility.
It is currently mapped to --abr 56 -mm, so that means that the recommendation would be to encode in mono, and use abr. As simple as that.

OK, haha, you're right, I can use anything I guess. But what would the recommended settings be for LAME 3.96.1 to be roughly equivalent to 3.97's -V2 setting? Thanks.


Thanks, that makes sense. I guess it's for statistical purposes only, and to me, the #s dont' mean much until I can compare it with other compressions.

Lame 3.96.1 also has the -V setting right? So use -V 2 or -V 3 (whatever fits your bit rate needs better) and be happy

This is quoted from the 'Quick Start' section of the "List of recommended LAME settings" thread:



Shouldn't there be a suggested setting for mono encoding in "High Quality" as well? There are good mono recordings that would be served by a HQ mono setting. From the recommendation above it sounds as if mono should be encoded with abr.

In the old days I think there was some kind of (albeit weak) consensus that -m m -b 80 could be added to the old --preset standard setting for mono encoding. Does this work OK with 3.97 as well?

-b80 is not necessary anymore with -vbr-new mode. But with the other VBR mode, I think that -b128 is still here by default. Therefore, -b80 should be worth.



EDIT: typo.

Thanks for the clarification!

In lame 3.97b does it store the setting used like the lame 3.90.3 modified compile did? In other words, if -V 2 = APS will the lameheader file indicate V2/APS was used somewhere?

On a diff note if V2 = APS on this compile then why do they get different bitrate avgs?

yes, this fuction was introduced in one of the recent lame releases and was backported to the old and unefficient Lame 3.90.3 release.

no, --preset standard and -V2 produces the exact same file (therefore same bitrate).

Hi everybody,

By default, ReplayGain analysis is enabled afaik it is not compatible with foobar's RG tag and portable mp3 players. So, wouldn't be better to add --noreplaygain to the command line to disable ReplayGain analysis and prevent LAME from applying gain to each track individually instead of to the album as a whole ?
Encoding a live or mixed album track by track, not as an image, with this command line option enabled isn't such a good idea, isn't it?

ReplayGain disabled: -V2 --vbr-new --noreplaygain

EAC's command line with replaygain disabled:


Please correct me if I'm wrong

Greetings

EDIT: typo

Please correct me if I'm wrong, but I thought LAME's replaygain analysis was for screen display only, during operation, and had nothing at all to do with applying gain to the tracks themselves. I have confirmed this to myself several times, using Mp3Gain.

That said, use of the --noreplaygain switch can still be useful. In my case at least, it speeds up the encoding process considerably.

Regards,
Madrigal

RG metadata is written to the LAME/INFO tag. The only program that I know of that uses that info is Otachan's in_mpg123.

It is also used by Madplay.

@kjoonlee & Gabriel:

Thanks for the clarification.

Regards,
Madrigal

Now that I'm using LAME 3.97 beta (despite my paranoia of beta software), I'm wondering if it's worth using --preset extreme (VBR) over --preset standard (VBR). In a double-blind test, is the slight difference in quality discernable by a good pair of ears? Thanks.

Great, Lame 3.97b2 has come out,
from changelog, it seems obvious to me, to recommend it instead of 3.97b1.

to make a quick answer, probably not, but you can check yourself.

for me v2 is transparent on all but a very few, extreme problem samples... i dont recall which ones at the moment, but in real music situations, i have yet to find a sample that i can abx.

For me V3 is almost transparent and it gives a good balance between quality and file size.

You're right. But I'd still like official recommendation from this forum before I upgrade the encoder. LAME 3.98 is also in its alpha 2 state, but who knows when it will hit beta.

IMHO, official recommendation should be updated to use LAME 3.97 beta 2 instead of beta 1 as the recommended LAME encoder version.

Gr.
Gonzalo

I PM'd "user" about it, so hopefully he'll update the first post of the thread and that someone will change the link from that 3rd sticky in this forum.

You can add Tag&Rename to the Tagging Software section...

Very good software!

http://www.softpointer.com/tr.htm