IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
For Dibrom (about his lame compiles)
RD
post Oct 1 2001, 14:24
Post #1





Group: Members
Posts: 151
Joined: 29-September 01
Member No.: 31



Dibrom I need some advice, please...

I have been putting off archiving my 500 cds for a long time.

I wanted to used lame 3.87, then waited for 3.88, then 3.89, and now am thinking of waiting for 3.90beta...

But Dibrom, do you really think there will be much difference in quality between 3.90beta --dm-preset insane and your current compile (alpha6) --dm-preset insane?

If the differences are minimal then I think I'll just use your compile now (alpha6) and start what I've been putting off for way too long...

Btw the way is your alpha 6 compile stable? or do you recommend i use the --dm-preset insane contained in lame 3.90 alpha7... this version of lame is bundled in win32lame1.7 and some say it is very stable for an alpha....but is it (do you think) more stable and better to use than your compile?

The thing that worries me about alpha7 is that although it held up well in the listening tests... I believe you or mp3fan has said that
--dm-highfreq was modified when renamed to --ns-sfb21 and hence may perform slightly differently?

So in the end I want to know if you think it is worth it to wait for the official lame3.90beta or to just go ahead with the alpha:
if I should go with the alpha--which one: your compile (alpha6) or the alpha7 compile.

Sorry for the length of the question,
I just want to finally archive these cds....its been so long now...
years....

RD
Go to the top of the page
+Quote Post
Dibrom
post Oct 1 2001, 22:03
Post #2


Founder


Group: Admin
Posts: 2958
Joined: 26-August 02
From: Nottingham, UK
Member No.: 1



There really shouldn't be a problem using the current lame alpha I don't think. I mean, its possible there may be further improvements down the line, but you are right, you can't wait forever smile.gif.

I'd suggest using --dm-preset standard though, or maybe xtreme. If you don't mind the higher bitrate of xtreme though, maybe --dm-preset standard -Z would be the best bet.

I hope that answers your questions smile.gif
Go to the top of the page
+Quote Post
RD
post Oct 1 2001, 23:51
Post #3





Group: Members
Posts: 151
Joined: 29-September 01
Member No.: 31



ok, i take you words to mean go with alpha7 then...

But you say " If you don't mind the higher bitrate of xtreme though, maybe --dm-preset standard -Z would be the best bet. "

Does this mean --dm-preset standar -Z has advantages over xtreme? and is larger than xtreme?

If so what advantages?

Thanks,
RD

PS looked at lame --longhelp |more
and -Z says toggles the scalefac feature on/off

adding -Z is increasing the bitrate but what is scalefac all about...
sorry last question then i'll (Try) to leave you alone about lame....
Go to the top of the page
+Quote Post
Dibrom
post Oct 2 2001, 06:31
Post #4


Founder


Group: Admin
Posts: 2958
Joined: 26-August 02
From: Nottingham, UK
Member No.: 1



QUOTE
Originally posted by RD
ok, i take you words to mean go with alpha7 then...

But you say " If you don't mind the higher bitrate of xtreme though, maybe --dm-preset standard -Z would be the best bet. "

Does this mean --dm-preset standar -Z has advantages over xtreme? and is larger than xtreme?

If so what advantages?

Thanks,
RD

PS looked at lame --longhelp |more
and -Z says toggles the scalefac feature on/off

adding -Z is increasing the bitrate but what is scalefac all about...
sorry last question then i'll (Try) to leave you alone about lame....


The thing is, the only real difference between --dm-preset standard and --dm-preset xtreme is that xtreme uses a lower ath curve. This effect probably isn't as noticeable if you don't have pretty good hearing above 16khz.

The original idea behind the presets were that "standard" was supposed to provide excellent quality at an acceptable bitrate in almost all situations. It addressed many quality issues, and it did attempt to reduce ringing, but ringing over 16khz was not addressed so much since this required an increased bitrate.

xtreme was supposed to provide a slightly lower ath, maybe a very slightly "cleaner" sound overall, though this would be hard to detect. The main difference is that frequencies over 16khz would be encoded a little bit more accurately at the sake of bitrate. Again, this isn't noticeable to most people, especially those that don't hear high frequency ringing in mp3 in the first place.

insane was, and still is, more of an experimental switch. It attempts to provide the absolute highest quality possible with LAME period. At the moment though, I'm not convinced it quite achieves this goal. I would recommend using standard instead for the most part.

Now, as for -Z.. the idea is that using noise shaping 1 instead of type will probably provide more of an audible improvement in quality, especially on critical samples, than that of just using a lower ath curve (which is what you would get from xtreme). I think in the end this makes more sense, and I'm getting ready to update the various presets with these ideas in mind.

What I think I may do:

1. Keep "standard" the way it is.
2. Modify xtreme to be the same as "standard" but with noise shaping 1 instead of 2.
3. Modify insane to the same as "standard" but with the lower athtype of the current xtreme, AND noise shaping 1. As well as possibly adding something like --ns-bass -8 and maybe -V1 or -V0

I'm not quite sure about the last part yet, I'll have to see..

Hope that helps.
Go to the top of the page
+Quote Post
Volcano
post Oct 3 2001, 16:12
Post #5





Group: Members (Donating)
Posts: 916
Joined: 30-September 01
From: Berlin, Germany
Member No.: 112



QUOTE
Originally posted by RD
Btw the way is your alpha 6 compile stable? or do you recommend i use the --dm-preset insane contained in lame 3.90 alpha7... this version of lame is bundled in win32lame1.7 and some say it is very stable for an alpha....but is it (do you think) more stable and better to use than your compile?

What about this LAME header? Correct me if I'm wrong, but I thought the later alpha versions write a corrupt header or something - r3mix recommended using the win32LAME 1.7 compile over the latest alphas (which is what I'm doing), as this version doesn't write the LAME header at all.

CU
Dominic
Go to the top of the page
+Quote Post
CiTay
post Oct 3 2001, 20:36
Post #6


Administrator


Group: Admin
Posts: 2378
Joined: 22-September 01
Member No.: 3



Later alphas don't write a corrupt header, they write an incomplete one (for instance, the ReplayGain data is missing). This has no effect whatsoever on sound quality, the latest alphas are safe to use. But you are right, the 3.90 alpha that comes with WinLAME 1.7 doesn't generate LAME tags.
Go to the top of the page
+Quote Post
Cloun
post Oct 4 2001, 22:03
Post #7





Group: Members
Posts: 1
Joined: 4-October 01
Member No.: 210



Hi Diprom,


where could I find which settings does the --dm-preset includes.

Thanx, Cloun
Go to the top of the page
+Quote Post
TrNSZ
post Oct 4 2001, 22:08
Post #8





Group: Developer
Posts: 717
Joined: 25-September 01
Member No.: 20



The best way to find the current settings for every preset at any moment is to use the source code browser. The settings are all included in a comment block before the code that actually enables the preset.

For example, surf over to:

http://cvs.sourceforge.net/cgi-bin/viewcvs...e.c?sortby=date

Click the latest revision's "Annonate" option to display the source. The revisions at the top is the most recent. If you go around line 787, you can see the dm-preset xtreme settings. Use your browsers search function from here on.
Go to the top of the page
+Quote Post
RD
post Oct 5 2001, 15:44
Post #9





Group: Members
Posts: 151
Joined: 29-September 01
Member No.: 31



Oh man, I'm sorry dibrom ... in my original post the first one before all these replies I meant to write about xtreme NOT insane... damn I didn't mean to confuse the issue and I did accidently.

What I really wanted to ask was:

"But Dibrom, do you really think there will be much difference in quality between 3.90beta --dm-preset xtreme and your current compile (alpha6) --dm-preset xtreme?
If the differences are minimal then I think I'll just use your compile now (alpha6) and start what I've been putting off for way too long... "

It was the xtreme preset that I was interested in....

Now while we are on the topic of xtreme:
(1) I take it that you do not regard xtreme as being as "experimental" as insane, correct? Xtreme, then, has proven itself to you, or do you still recommend the use of standard over xtreme?

(2) what is better:
--------------------------
(A) standard with -Z
or
(B) xtreme without -Z

Thanks,
and sorry for the confusion....
I could kick myself for the mistake...

RD
Go to the top of the page
+Quote Post
Dibrom
post Oct 5 2001, 16:12
Post #10


Founder


Group: Admin
Posts: 2958
Joined: 26-August 02
From: Nottingham, UK
Member No.: 1



QUOTE
Originally posted by Cloun
Hi Diprom,

where could I find which settings does the --dm-preset includes.

Thanx, Cloun


At the moment I am reworking the xtreme and insane presets, I'd rather wait until I'm finished before getting into a discussion about the details of the switches, though you can find the current ones from the links TrNSZ provided

QUOTE
Originally posted by RD
It was the xtreme preset that I was interested in.... 

Now while we are on the topic of xtreme: 
(1) I take it that you do not regard xtreme as being as "experimental" as insane, correct? Xtreme, then, has proven itself to you, or do you still recommend the use of standard over xtreme? 

(2) what is better: 
-------------------------- 
(A) standard with -Z 
or 
(B) xtreme without -Z


Both standard and xtreme have stayed exactly the same as the original compile. There is a very slight difference in the implementation of --ns-sfb21 vs that of --dm-highfreq, but that is only so that --ns-sfb21 can work in addition to --ns-treble, whereas before --dm-highfreq would completely override that switch. There should be no quality difference.

1. Yes, xtreme is not really experimental, but only slightly better than standard. I am changing this switch around at the moment (working on a new compile) though to offer a larger margin of higher quality than standard, especially on more critical samples.

2. standard with -Z, but again this may even change as well once I modify the switches some.
Go to the top of the page
+Quote Post
RD
post Oct 5 2001, 20:21
Post #11





Group: Members
Posts: 151
Joined: 29-September 01
Member No.: 31



Thanks to everybody for their help,
and a special thanks to Dibrom--you are the best!

I wish I could pour you a large glass of Chimay blue label!!
:drink:
Go to the top of the page
+Quote Post
deranger
post Oct 5 2001, 21:05
Post #12





Group: Members
Posts: 27
Joined: 5-October 01
Member No.: 221



I've just read in r3mix's forum:

QUOTE
Posted by r3mix: If you upsample a 44.1kHz to 48kHz then for the same "amount" of music, there will be more samples, which will allow the encoder to encode with a 9% shorter eacho reaction.  So at any decent (let's assume +128kbit/s) bitrate upsampling to 48kHz will lower the chance of echo problems with (48000/44100-1)*100 % but given only a very few people already heard it it will make both a very significant difference to those and it'll fix the echo problem altogether

Posted by TrNSZ: It's best to use the SSRC resampler to resample to 48/24 without dithering before encoding to MP3. 

I have a batch file (lamefr.cmd) which I use as a front-end for LAME in CDex or EAC.  It currently contains:

@ECHO OFF
ssrc --rate 48000 --bits 24 --twopass --dither 0 %1 C:Tempssrc.wav
lame390a7 --dm-preset xtreme --scale 0.96 C:Tempssrc.wav %2
del c:tempssrc.wav 
 


Do you advise doing such a resampling ? Would it really increase the quality of my mp3s ?!
Go to the top of the page
+Quote Post
TrNSZ
post Oct 6 2001, 19:54
Post #13





Group: Developer
Posts: 717
Joined: 25-September 01
Member No.: 20



There are pro's and con's to this 48khz resample before encode.

The good points are that, yes, you will definately improve quality. If you have an AC97/PC99 complaint soundcard, by making 48kHz MP3's you avoid the soundcard and Windows kmixer resampler, which makes the music sound less lively and adds noise when converting on-the-fly from 44.1kHz to 48kHz. Also, when using 48kHz, you reduce pre- and post-echo artifacts, which really helps on samples like castanets.

The bad side of using 48kHz MP3's is that you have to make sure your portable device, if you use one, works properly with them. Also, if you plan to burn Audio CD's from the MP3's, you lose simple drag and drop burning, because all CD-R software (other than Nero) uses the Windows kmixer to resample 48kHz to 44.1kHz, which results in terrible quality. Nero resamples using a quality algorithm, but the CD's are burned in mono mode only, not stereo, because of a bug that has yet to be fixed.
Go to the top of the page
+Quote Post

Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 29th December 2014 - 13:00