Hello!

I have a kind of awkward problem regarding MP3s that I ripped myself. I used Lame 3.97 and the recommended settings from hydrogenaudio.

I did some ABX-testing today in order to see how good ogg is in comparison with Lame which I always prefered for ripping. Well I was rather shocked when I realized that I could easily almost on all samples could ABX the Lame file when compared to the original wav file. That was something which I didn't expect.

On the other hand I could ABX Aotuvs Beta5 at q5 also but it was more difficult. I didn't manage to ABX q6 on the Samples I used.

I am a little bit scared now as my hole music collection is ripped with Lame in the recommended settings and I don't know if I should go for Ogg now as Lame doesn't seem to improve above V2 settings.

I can't go for lossless as my Laptop drive isn't big enough for that (I own around 450 CDs I guess). I do have a flash based player that allows me to play ogg, mp3 and wma files and I don't want to store music in different formats and/or filesizes for Laptop and flash player.

Any recommendations for me?

Thank you!

What are the exact artifacts you're noticing? If they regard known problem points with MP3 - particularly with lowpass and preecho, I guess - moving to a well-maintained codec where those issues are less of a concern should be a safe move.

Mainly I notice them on drums (hihats). I mostly listen to rock music and it seemed kind of annoying to me here.

But I also noticed that the whole song seemed to sound a bit different in the heights that the original. How should I describe that?! It seemed to have more emphasis on the heights that the original ... I know it sounds strange ...

Lame 3.97 has it's issues though -V2 is expected to be transparent usually.
Anyway with 3.98beta the Lame devs have made a big progress. The official 3.98b6 has just been released, and you're welcome to try it. Quality scales well, so if you shouldn't be satisfied by -V2 it's worth while trying -V1 or -V0.

I will have a closer look on 3.98b6 then and do some more testing to find the right setting for my purpose.

I will also take aoTuVb5 into account. My iRiver T60 seems to handle them well ... my old iMP-550 had real issues here.

Edit: I just noticed a problem with 3.98b6. I use the dll-Version with CDex but it always gives me the same filesize, even if I tell the program to use V0 or V1.

How sure are you that you had been encoding at -V2?

What are you using for ABX testing? Could you provide some sample logs?

What's your settings??
Try these:

Version: MPEG I
Bitrate Min: 32
Bitrate Max: 320
Mode: J-Stereo
Quality: High(q=2)
VBR Method: VBR-New
VBR Quality: V n [where n is 2 or anything you want]
Output samplerate: Auto

CDex produces exactly the same result as "lame.exe -V n --vbr-new file.wav"

why not just use command line encoding for a few test files so you know exactly what you are getting?

and i agree with what has been said, if v2 isn't good enough try v1, or v0 or 320, or else move to ogg.

I used Foobar for Testing and the file information told me I encoded at -V2.

I will do some more testing. Are there any other programs recommended for ABX-Testing? I will post some logs then.

I just have a log from a short test I did. I stopped after 6 tries with this test as it seemed pretty obvious with this special song.

foo_abx 1.3.1 report
foobar2000 v0.9.4.5
2007/12/22 16:07:57

File A: C:\Dokumente und Einstellungen\Thomas\Eigene Dateien\Eigene Musik\Various\ROCK HARD FESTIVAL 2007 - LIVE -\01-ARMORED SAINT - March Of The Saint.mp3
File B: C:\Dokumente und Einstellungen\Thomas\Eigene Dateien\Eigene Musik\Various\ROCK HARD FESTIVAL 2007 - LIVE -\01-ARMORED SAINT - March Of The Saint.wav

16:07:57 : Test started.
16:09:27 : 00/01 100.0%
16:10:16 : 01/02 75.0%
16:10:37 : 02/03 50.0%
16:11:39 : 03/04 31.3%
16:12:24 : 04/05 18.8%
16:13:18 : 05/06 10.9%
16:13:25 : Test finished.

----------
Total: 5/6 (10.9%)

you may want to try reaching 0.0% coz with 10% you still could be guessing... Is this 3.98b6 V2? Could you ABX at V0.... Many things I can ABX at V2, I usually cannot at V0. And this newest lame version ABX is even more difficult... like I said in another topic, LAME is reaching the point of becoming a very refined codec and compete well even with Vorbis in which you could not distinguish which one is each.

0% can never be reached. What you mean is something "closer" to 0%, not 0% itself.

does your laptop have a dvd burner? if it does, you can burn the lossless files to dvd and keep the oggs since your player supports them. if you have more cash, you could get an external hd.

just being pratical lyx...

from topic: http://www.hydrogenaudio.org/forums/index....c=58152&hl=

Are you using -V2 --vbr-new or -V2 --vbr-old ??
If you use foobar for encoding...
you should use "Fast mode" with it.

I was using LAME 3.97 -V2 --vbr-new.

However I just tested the same sample with LAME 3.98b6 -V2 --vbr-new and it seemed that it improved quite a lot for me!



foo_abx 1.3.1 report
foobar2000 v0.9.4.5
2007/12/23 08:22:48

File A: C:\Dokumente und Einstellungen\Thomas\Eigene Dateien\Eigene Musik\Various\Rock Hard Festival 2007\01-ARMORED SAINT - March Of The Saint.mp3
File B: C:\Dokumente und Einstellungen\Thomas\Eigene Dateien\Eigene Musik\Various\Rock Hard Festival 2007\01-ARMORED SAINT - March Of The Saint.wav

08:22:48 : Test started.
08:23:44 : 01/01 50.0%
08:25:11 : 01/02 75.0%
08:25:56 : 02/03 50.0%
08:26:54 : 02/04 68.8%
08:29:20 : 02/05 81.3%
08:30:10 : 03/06 65.6%
08:30:32 : 03/07 77.3%
08:30:58 : 03/08 85.5%
08:31:55 : 04/09 74.6%
08:32:34 : 05/10 62.3%
08:33:56 : 05/11 72.6%
08:34:06 : Test finished.

----------
Total: 5/11 (72.6%)


The anoying sounds I heard regarding the hihats are totally gone here.

For comparison here the results for aoTuVb5 -q6:


foo_abx 1.3.1 report
foobar2000 v0.9.4.5
2007/12/23 08:40:43

File A: C:\Dokumente und Einstellungen\Thomas\Eigene Dateien\Eigene Musik\Various\Rock Hard Festival 2007\01-ARMORED SAINT - March Of The Saint.wav
File B: C:\Dokumente und Einstellungen\Thomas\Eigene Dateien\Eigene Musik\OGG\Various\Rock Hard Festival 2007\01-ARMORED SAINT - March Of The Saint.ogg

08:40:43 : Test started.
08:41:48 : 01/01 50.0%
08:42:47 : 01/02 75.0%
08:43:30 : 02/03 50.0%
08:44:10 : 02/04 68.8%
08:44:33 : 03/05 50.0%
08:45:15 : 03/06 65.6%
08:46:33 : 03/07 77.3%
08:46:49 : 04/08 63.7%
08:47:00 : 05/09 50.0%
08:47:32 : 06/10 37.7%
08:48:06 : 06/11 50.0%
08:48:47 : 06/12 61.3%
08:49:04 : 07/13 50.0%
08:49:32 : 08/14 39.5%
08:49:59 : 09/15 30.4%
08:50:15 : 10/16 22.7%
08:51:22 : 11/17 16.6%
08:52:51 : 11/18 24.0%
08:53:18 : 11/19 32.4%
08:53:32 : 11/20 41.2%
08:53:45 : 11/21 50.0%
08:54:11 : 12/22 41.6%
08:54:34 : 12/23 50.0%
08:54:38 : Test finished.

----------
Total: 12/23 (50.0%)

-V2 sounds transparent to me. I'm sorry if this is the wrong place (I'm very new around here), but I want to use -V2 with a minimum bitrate of 160. Are these the right settings? Would these be of high enough quality for just listening to music using cheap speakers and an iPod?

My settings are:

-b 160 -V2 --vbr-new -q0

(I use EAC to rip my CDs)

Setting a lower bitrate limit does not increase audio quality. Setting an upper limit would decrease audio quality.
If you have a flawed mp3 device, then it would make some sense, but then you would like to enforce the minimum bitrate for silence too: "-V2 --vbr-new -b160 -F"

Thanks for the info. And seeing as I am re-ripping my whole CD collection (due to previously ripping them @ 128 kbps ), do you recommend I use:

'-V2 --vbr-new -q0'?

I would suggest lame 3.98 beta 6 with a plain "-V2" setup.

OK, thanks again.

I also did a quick test with a song from my Alien Ant Farm CD, and noticed even with the '-b 160', it didn't use anything lower than 192 kbps (apart from 32 kbps, which I assume is for complete silence).

Edit: Oh, and by the way, why do you recommend 3.98 beta 6? (I am currently using the final version of 3.97)

I guess it was me who first suggested to the OP to give 3.98b6 a try.
There are some reasons to do so

• the extremely bad pre-echo sample 'eig' has improved a lot with 3.98b6. So we can expect this improvement to be relevant for more pre-echo problems. The OP wrote about issues with drums, so 3.98b6 is a candidate for the solution of these problems. Meanwhile diewaldo has found it is the solution.
• 3.97 has a specific problem called 'sandpaper noise' which affects voices. This problem is gone with 3.98b6. This kind of problem can be heard for instance with the 'Birds' sample.
• There were certain weaknesses with Lame's psy model not just with 3.97, but also with 3.96 and other versions, and it was especially bad when using VBR (as VBR enhances the goods and bads of the psy model). These problems are gone with 3.98b6. My top sample for these kind of problems is 'trumpet' (before I heard a 3.97 -V2 encoding of trumpet the mp3 world was fine to me, but with this sample I was totally upset. With 3.98 the world is fine again). I'm not sure whether or not the 'sandpaper noise problem' also falls into this category - from mere listening the problem sounds similar.

The HA recommendations are just a guideline, and only final versions will make it into it.
With 3.98b6 we're certainly so close to final (qualitywise) that in my very personal opinion it would be unwise to prefer 3.97 final just for the academic sake of using a final version.

Sounds good, but I think I'm going to wait a while until 3.98 final comes out.

Although I might do some quick listening tests tomorrow just to see how it works out.

Why do you advise against -q0?
Does it make quality worse, or does it just take more encoding time than it's worth? or something else?

-V0 is the highest VBR setting you can have, making larger files. Unless you have golden ears, usually you will not need such high bit rates for your audio to sound transparent.

Q0 is slower, but of higher quality, and large file size.

Did I? Just try LAME 3.98b6 with plain "-V2".


jamesbaud talked about -q0, not -V0.

You seemed to. Maybe I'm mistaken. Here, the original poster said:



You said:



So it seems like you don't think -q0 is a good idea. If so, is the fact that -q0 is slower and causes larger files (according to SamHain86, below) the reason you would recommend against -q0? Or is there some other reason?



By the way, thank you, Robert, and all the lame devs. I hope you all have a great holiday and I wish that 2008 brings you love, prosperity, and lots of inspiration for improving lame!

jb

LAME 3.98 defaults to -q0, you'd be wasting 3 button pressings of your time

I thought -qs 0 through 2 were identical in vbr-new?

For 3.98 beta6 and vbr-new -- q0 through q4 are identical .

added: lame build from www.rarewares.org

After I read this, I decided to run a couple of test encodings on the most recent 3.98b6 from lame.bakerweb.biz to see if I could "save" button pressings.

Here's my original command line:

-V 2 --vbr-new -q 0 --strictly-enforce-ISO --noreplaygain

Here's my modified one:

-V 2 --strictly-enforce-ISO --noreplaygain

I used the same file and encoded using both command lines. I compared the files in Mr Questionman and LameTag. There was no difference in file sizes or quality according to each program.

What is interesting is that I tried to run the same line without --strictly-enforce-ISO and found that there WAS a file size difference. (without adding the ISO, the file size was smaller) I thought that --strictly-enforce-ISO was enabled by default. I guess I can omit "--vbr-new -q 0 " but that's about all the button pressings I can "save."

For me the new LAME 3.98b6 is topping AoTuvb5! Some proof:

LAME -vbr-new -V2

foo_abx 1.3.1 report
foobar2000 v0.9.4.5
2007/12/26 20:11:26

File A: C:\Dokumente und Einstellungen\Thomas\Eigene Dateien\Eigene Musik\Sebastian Bach\Angel Down LAME -V2\02-You Don't Understand.mp3
File B: C:\Dokumente und Einstellungen\Thomas\Eigene Dateien\Eigene Musik\Sebastian Bach\Angel Down LAME -V2\02-You Don't Understand.wav

20:11:26 : Test started.
20:12:05 : 00/01 100.0%
20:12:35 : 00/02 100.0%
20:13:04 : 00/03 100.0%
20:13:53 : 01/04 93.8%
20:14:04 : 02/05 81.3%
20:14:20 : 03/06 65.6%
20:14:41 : 03/07 77.3%
20:15:01 : 03/08 85.5%
20:15:18 : 04/09 74.6%
20:15:42 : 05/10 62.3%
20:16:04 : 06/11 50.0%
20:17:17 : 06/12 61.3%
20:17:38 : 07/13 50.0%
20:17:51 : 08/14 39.5%
20:18:13 : 09/15 30.4%
20:18:32 : 09/16 40.2%
20:19:01 : 09/17 50.0%
20:19:22 : 10/18 40.7%
20:19:40 : 11/19 32.4%
20:20:07 : 12/20 25.2%
20:20:13 : Test finished.

----------
Total: 12/20 (25.2%)



AoTuv -q6:

foo_abx 1.3.1 report
foobar2000 v0.9.4.5
2007/12/26 20:21:26

File A: C:\Dokumente und Einstellungen\Thomas\Eigene Dateien\Eigene Musik\Sebastian Bach\Angel Down AoTuvb5 q6\02-You Don't Understand.ogg
File B: C:\Dokumente und Einstellungen\Thomas\Eigene Dateien\Eigene Musik\Sebastian Bach\Angel Down AoTuvb5 q6\02-You Don't Understand.wav

20:21:26 : Test started.
20:21:47 : 01/01 50.0%
20:22:17 : 02/02 25.0%
20:23:03 : 03/03 12.5%
20:23:42 : 04/04 6.3%
20:24:24 : 05/05 3.1%
20:24:31 : Test finished.

----------
Total: 5/5 (3.1%)



AoTuv -q7:

foo_abx 1.3.1 report
foobar2000 v0.9.4.5
2007/12/26 20:32:06

File A: C:\Dokumente und Einstellungen\Thomas\Eigene Dateien\Eigene Musik\Sebastian Bach\Angel Down AoTuvb5 q7\02-You Don't Understand.ogg
File B: C:\Dokumente und Einstellungen\Thomas\Eigene Dateien\Eigene Musik\Sebastian Bach\Angel Down AoTuvb5 q7\02-You Don't Understand.wav

20:32:06 : Test started.
20:32:18 : 01/01 50.0%
20:32:30 : 01/02 75.0%
20:32:43 : 01/03 87.5%
20:33:10 : 02/04 68.8%
20:33:53 : 02/05 81.3%
20:34:26 : 02/06 89.1%
20:34:48 : 03/07 77.3%
20:34:55 : 04/08 63.7%
20:35:03 : 05/09 50.0%
20:35:10 : 06/10 37.7%
20:35:22 : 07/11 27.4%
20:35:41 : 08/12 19.4%
20:36:11 : 09/13 13.3%
20:36:30 : 10/14 9.0%
20:36:35 : Test finished.

----------
Total: 10/14 (9.0%)


For my ears something with ogg is fundamentally wrong! The sound is lacking heights ... it is kind of muffled.

I did some more testing today and the outcome is completely different from yesterday. How is this possible? I am a bit confused right now.

LAME -vbr-new V2:

foo_abx 1.3.1 report
foobar2000 v0.9.4.5
2007/12/27 19:53:11

File A: C:\Dokumente und Einstellungen\Thomas\Eigene Dateien\Eigene Musik\Danzig\Danzig\01-Twist Of Cain.wav
File B: C:\Dokumente und Einstellungen\Thomas\Eigene Dateien\Eigene Musik\Danzig\Danzig\01-Twist Of Cain.mp3

19:53:11 : Test started.
19:53:44 : 01/01 50.0%
19:54:12 : 02/02 25.0%
19:54:38 : 02/03 50.0%
19:54:47 : 02/04 68.8%
19:55:01 : 03/05 50.0%
19:55:31 : 04/06 34.4%
19:55:44 : 04/07 50.0%
19:55:59 : 05/08 36.3%
19:56:15 : 06/09 25.4%
19:56:27 : 07/10 17.2%
19:56:42 : 08/11 11.3%
19:57:07 : 09/12 7.3%
19:57:23 : 10/13 4.6%
19:57:41 : 11/14 2.9%
19:57:57 : 12/15 1.8%
19:58:07 : 13/16 1.1%
19:58:20 : 14/17 0.6%
19:58:41 : 15/18 0.4%
19:59:06 : 16/19 0.2%
19:59:18 : 17/20 0.1%
19:59:31 : 18/21 0.1%
19:59:40 : 19/22 0.0%
19:59:44 : Test finished.

----------
Total: 19/22 (0.0%)



AoTuVb5 -q7:

foo_abx 1.3.1 report
foobar2000 v0.9.4.5
2007/12/27 19:43:26

File A: C:\Dokumente und Einstellungen\Thomas\Eigene Dateien\Eigene Musik\Danzig\Danzig\01-Twist Of Cain.ogg
File B: C:\Dokumente und Einstellungen\Thomas\Eigene Dateien\Eigene Musik\Danzig\Danzig\01-Twist Of Cain.wav

19:43:26 : Test started.
19:45:36 : 00/01 100.0%
19:46:16 : 01/02 75.0%
19:46:35 : 02/03 50.0%
19:47:07 : 02/04 68.8%
19:47:46 : 02/05 81.3%
19:47:54 : 03/06 65.6%
19:48:02 : 04/07 50.0%
19:48:14 : 04/08 63.7%
19:48:24 : 05/09 50.0%
19:48:34 : 05/10 62.3%
19:48:59 : 05/11 72.6%
19:49:30 : 05/12 80.6%
19:49:51 : 06/13 70.9%
19:50:05 : 07/14 60.5%
19:50:12 : 08/15 50.0%
19:50:20 : 09/16 40.2%
19:50:31 : 10/17 31.5%
19:50:57 : 10/18 40.7%
19:51:10 : 10/19 50.0%
19:51:24 : 11/20 41.2%
19:51:42 : 11/21 50.0%
19:51:57 : 12/22 41.6%
19:52:09 : 13/23 33.9%
19:52:18 : 13/24 41.9%
19:52:33 : 14/25 34.5%
19:52:35 : Test finished.

----------
Total: 14/25 (34.5%)


AoTuV -q6:

foo_abx 1.3.1 report
foobar2000 v0.9.4.5
2007/12/27 20:34:44

File A: C:\Dokumente und Einstellungen\Thomas\Eigene Dateien\Eigene Musik\Danzig\Danzig AoTuVb5 q6\01-Twist Of Cain.ogg
File B: C:\Dokumente und Einstellungen\Thomas\Eigene Dateien\Eigene Musik\Danzig\Danzig wav\01-Twist Of Cain.wav

20:34:44 : Test started.
20:35:19 : 00/01 100.0%
20:36:09 : 01/02 75.0%
20:36:50 : 01/03 87.5%
20:37:20 : 01/04 93.8%
20:37:46 : 02/05 81.3%
20:38:03 : 03/06 65.6%
20:38:19 : 03/07 77.3%
20:38:30 : 04/08 63.7%
20:38:48 : 04/09 74.6%
20:39:04 : 04/10 82.8%
20:39:28 : 04/11 88.7%
20:39:40 : 04/12 92.7%
20:39:49 : 04/13 95.4%
20:40:18 : 05/14 91.0%
20:40:28 : 05/15 94.1%
20:40:45 : 06/16 89.5%
20:40:58 : 07/17 83.4%
20:41:25 : 08/18 76.0%
20:41:45 : 09/19 67.6%
20:42:02 : 09/20 74.8%
20:42:16 : 09/21 80.8%
20:42:42 : 10/22 73.8%
20:42:59 : 11/23 66.1%
20:43:17 : 12/24 58.1%
20:43:41 : 13/25 50.0%
20:43:58 : 14/26 42.3%
20:44:12 : 15/27 35.1%
20:44:24 : 15/28 42.5%
20:44:56 : 16/29 35.6%
20:45:06 : 17/30 29.2%
20:45:19 : 17/31 36.0%
20:45:37 : 17/32 43.0%
20:45:50 : 18/33 36.4%
20:45:55 : Test finished.

----------
Total: 18/33 (36.4%)



None of them seems perfect. But here LAME is giving less Bass than the wav. There is less pressure in the music.

Not all songs are the same so not all encoding is the same. Lame at -V 2 might be transparent for songs x, y, and z but that doesn't mean that is is transparent for song a, b, and c. Also your ears change over night. You can spend all day training them to hear these audio flaws then go to bed. When you wake up your ears will be different until you start training them again.

I have been looking at your ABX results and it seems that Lame at -V 2 is more than enough for you. You early test results were hard to tell as you only went through 5 times. Your later test results (except for that one) seem to point that you should be fine with -V 2 encoding. All formats have their killer samples or instruments/sounds that give them trouble. The iTunes AAC encoder has trouble with thunder no matter what bitrate you encode it at. I would do just one more ABX test. Pick 1-5 songs that represent the majority of music in your collection and run your tests. You don't want to choose classical music if you only have one or two albums compared to 100 rock albums.

OK, I think I begin to understand.

Now I am asking myself if all codecs have their flaws if it might be an option to choose a lower setting. If I don't compare with the wav even ogg -q5 is sounding good.

Perhaps I should not make the tests by opting for a higher bitrate in order to get rid of the flaws that I can't get rid of and should instead look down the ladder and compare what bitrate is giving me the best result for a lesser bitrate.

When listening to an imperfect encoding to me the key question is: is it annoying within practical listening?
If the answer is no, you can stick with -V2, or allow for a lower quality setting as long as you get fine results in an overall sense.
If the answer is yes, you can try a higher quality setting like -V0 or CBR 320. You do get a chance that things will improve. If you are still not content even with very high bitrate you can switch encoder or codec and find the one that matches your needs best.

I was listening today Information Society 1988 album which is very electronic and has got a lot of cymbals.

I heard a 128kbps AAC encoding, and a LAME 3.90 --alt-preset-standard encoding (V2) I was so surprised of how much smearing and distortion the MP3 file had. AAC even at 128kbps had a little distortion but less than MP3. This is a scenario in which you clearly see some benefits of a modern codec. (Some claim that 3.90 is still best than 3.97).

Do you mind trying Lame 3.98b6 on your album as it seems to have improved a lot on temporal resolution problems (and more)?

I don't have the lossless sources, the AAC files are from iTunes, and the MP3 file is from the internet, the MP3 has been ripped properly according to the log.

In this case you can't compare your aac and mp3 encodings as you don't know nothing about the history especially of the mp3 encoding. Maybe it was re-encoded.
Don't let me be misunderstood: I can imagine well iTunesAAC@128kbps encodings outperforming 3.90.3@aps encodings, but your samples aren't a real basis for showing this up.
What's more important: when talking about Lame we shouldn't miss 3.98b6 as this new version is really a big improvement in many respect.

After trying out a little bit more I decided to go with AoTuVb5 -q5 for the future. It still sounds good enough with my iriver T60 and it is also sufficient for my laptop. Also I get some more free space on the harddrive when compared to Lame -v2 vbr-new.

Thank you all for your help!

I think I won't need lossless as I already have my CDs. OK, it might be more convenient if you have them on your harddrive, but you won't see the Artwork either. I personally like to read booklets while listening to the songs.

I realized I might even be suited well with AoTuVb5 -q4 but I like beeing "on the safe side".

Have a Happy New Year 2008!

compromise: -q4.5