Nero AAC Codec 1.3.3.0 Options

[Brent]

Thanks for posting that link sketchy_c, I updated the first post.

Any comment on why the Linux build is slower than the Windows build in Wine?

[lvqcl]

Any comment on why the Linux build is slower than the Windows build in Wine?

Just a guess: Windows version was built with Intel compiler (and it uses IPP - intel performance primitives), and Linux version was built with GCC?

[menno]

Hmm, I'm experiencing a problem playing back 6 (5.1) channel sound with libfaad2 (in the latest ffdshow) and the built in AAC decoder in Media Player Classic encoded with the new 1.3.3.0 codec. Everything has an odd underwater echo.

The problem doesn't occur with the latest version of Winamp and in_mp4.dll

The problem doesn't occur with any of the above programs with 2 channel.

See http://pcdom.com.au/neroaac/ for examples.

Thanks,

Dom.

This seems to be a problem with those players specifically, probably downmixing incorrectly or wrong channel ordering. I tried latest FAAD2 libraries and it sounds fine. It could be an old bug in FAAD2 too, but only if those players still use very old versions of it (which is very possible).

[Sylph]

So no one knows why the customizations were dropped and why there's no LAME plugin for Nero (used to be there for version 8)?

[muaddib]

So no one knows why the customizations were dropped and why there's no LAME plugin for Nero (used to be there for version 8)?

New improved Lame and MP4 plugins are there in Nero9. You will have to use "Encode Files" dialog to access them.

[Sylph]

So no one knows why the customizations were dropped and why there's no LAME plugin for Nero (used to be there for version 8)?

New improved Lame and MP4 plugins are there in Nero9. You will have to use "Encode Files" dialog to access them.

OK, but with that dialogue I can't rip a CD, or can I? Would I first have to use Save Tracks as WAVs and then encode them or...?

[muaddib]

New improved Lame and MP4 plugins are there in Nero9. You will have to use "Encode Files" dialog to access them.

OK, but with that dialogue I can't rip a CD, or can I? Would I first have to use Save Tracks as WAVs and then encode them or...?

Yes, you are right. You first have to save tracks to WAVs. But I believe that this is topic for a separate thread where it could be further discussed.

[singaiya]

I encoded a few albums with the new codec using my old settings of -q 0.25 lc, which forces LC and usually averages about 100 kbps. It's always been transparent like that but right away I suspected something was wrong because now there was swishy cymbals. Maybe it's just when forcing LC -- I haven't tried not using that switch yet. But the reason I use it is because I've noticed that when I try to average 100 kbps many times it picks HE profile to use and I don't want that (ipod).

As you can see in the log, I was sure of my choice in each trial within seconds -- it wasn't even hard and it's basic rock and roll not a known killer sample. It's the absolute first thing I happened to encode and listen to with this encoder. And the rest of the album sounds that bad too with this setting.

This song averages 103 kbps. Edit: It is 65 kbps. The -q values have been remapped since the Feb 2007 encoder two versions ago, which is what I have been using all this time. See my next post for tests that produce 103 kbps.

CODE

foo_abx 1.3.3 report
foobar2000 v0.9.5.2
2008/10/04 21:36:14

File A: C:\Documents and Settings\All Users\Documents\Young Fresh Fellows\[] When We Were Good\16. Searchin' U.S.A..lossy.tak
File B: C:\Documents and Settings\All Users\Documents\Young Fresh Fellows\Topsy Turvy\01. Searchin' U.S.A..m4a

21:36:14 : Test started.
21:36:59 : 01/01  50.0%
21:37:10 : 02/02  25.0%
21:37:20 : 03/03  12.5%
21:37:25 : 04/04  6.3%
21:37:31 : 05/05  3.1%
21:37:38 : 06/06  1.6%
21:37:47 : 07/07  0.8%
21:37:51 : 08/08  0.4%
21:37:57 : Test finished.

----------
Total: 8/8 (0.4%)

.
edit: just to verify, I also did an abx with my older version (feb nero) and had no hope abx'ing it (1/3 and I gave up because the 1 was a lucky guess anyway). Could someone else please try any random song at that setting on the new encoder to get another statistic?

This post has been edited by singaiya: Oct 5 2008, 21:08

[melomaniac]

Could you try q 0.35 please? I also have an iPod and I've never had a HE file with that preset (bitrate ~100-105 kbps).

[zombiewerewolf]

Here is my ABX result of Nero -q 0.25 -lc, a quick test, only one song.

Difference between Original and encoded files is pretty obvious.
Difference between 1.1.34.2 and 1.3.3.0 is more subtle but still noticeable.

Personally I prefer sound produced by version 1.3.3.0.
I feel that 1.1.34.2 produces a little bit like under-water kind of sound.

Simple Plan - 06 Generation (NERO 1.1.34.2).mp4 = 102 Kbps, 2,360,294 bytes
Simple Plan - 06 Generation (NERO 1.3.3.0).mp4 = 84.6 Kbps 1,967,167 bytes

CODE

ABC/HR Version 1.1 beta 2, 18 June 2004
Testname: Nero -q 0.25 LC

1R = C:\!TEMP\Listening Test\Simple Plan - 06 Generation (NERO 1.3.3.0).wav
2R = C:\!TEMP\Listening Test\Simple Plan - 06 Generation (NERO 1.1.34.2).wav

---------------------------------------
General Comments:

---------------------------------------
1R File: C:\!TEMP\Listening Test\Simple Plan - 06 Generation (NERO 1.3.3.0).wav
1R Rating: 4.5
1R Comment:
---------------------------------------
2R File: C:\!TEMP\Listening Test\Simple Plan - 06 Generation (NERO 1.1.34.2).wav
2R Rating: 4.0
2R Comment:
---------------------------------------
ABX Results:
Original vs C:\!TEMP\Listening Test\Simple Plan - 06 Generation (NERO 1.3.3.0).wav
10 out of 10, pval < 0.001
Original vs C:\!TEMP\Listening Test\Simple Plan - 06 Generation (NERO 1.1.34.2).wav
10 out of 10, pval < 0.001
C:\!TEMP\Listening Test\Simple Plan - 06 Generation (NERO 1.3.3.0).wav vs C:\!TEMP\Listening Test\Simple Plan - 06 Generation (NERO 1.1.34.2).wav
19 out of 21, pval < 0.001

[singaiya]

More news: one thing to clear up about the awful sound of my last test - apparently the bitrate of the song I tested at 0.25 LC was 65 kbps. When I wrote 103, that is what the older Feb version produces with this setting, which I failed to ABX. So at 65 kbps of course it will sound awful.

Then I tried 1.3.3.0 at both 0.35 LC and 0.35 without LC, from your suggestion, melomaniac. This time both came out at 103 kbps but still easily ABX-able. I will continue to use the older Feb encoder.

Nero 1.3.3.0 -q 0.35 -LC (103 kbps):

CODE

foo_abx 1.3.3 report
foobar2000 v0.9.5.2
2008/10/05 10:53:49

File A: C:\Documents and Settings\All Users\Documents\Young Fresh Fellows\[] When We Were Good\16. Searchin' U.S.A..lossy.tak
File B: C:\Documents and Settings\All Users\Documents\Young Fresh Fellows\[] When We Were Good\Young Fresh Fellows\Topsy Turvy\01. Searchin' U.S.A..m4a

10:53:49 : Test started.
10:54:51 : 01/01  50.0%
10:55:05 : 02/02  25.0%
10:55:15 : 03/03  12.5%
10:55:31 : 04/04  6.3%
10:55:42 : 05/05  3.1%
10:55:57 : 06/06  1.6%
10:56:09 : 07/07  0.8%
10:56:27 : 08/08  0.4%
10:56:31 : Test finished.

----------
Total: 8/8 (0.4%)

Nero 1.3.3.0 -q 0.35 (no forcing -LC, still 103 kbps):

CODE

foo_abx 1.3.3 report
foobar2000 v0.9.5.2
2008/10/05 10:58:57

File A: C:\Documents and Settings\All Users\Documents\Young Fresh Fellows\[] When We Were Good\16. Searchin' U.S.A..lossy.tak
File B: C:\Documents and Settings\All Users\Documents\Young Fresh Fellows\[] When We Were Good\16. Searchin' U.S.A..lossy.m4a

10:58:57 : Test started.
10:59:30 : 01/01  50.0%
10:59:49 : 02/02  25.0%
11:00:09 : 03/03  12.5%
11:00:28 : 04/04  6.3%
11:00:49 : 05/05  3.1%
11:00:59 : 06/06  1.6%
11:01:08 : 07/07  0.8%
11:01:22 : 08/08  0.4%
11:01:24 : Test finished.

----------
Total: 8/8 (0.4%)

[melomaniac]

Maybe you should post your sample. See if anyone else could detect this problem at upper bitrates.

This post has been edited by melomaniac: Nov 2 2008, 12:34

[lvqcl]

Nero 1.3.3.0 -q 0.35 (no forcing -LC, still 103 kbps):

NeroAacEnc 1.3.3.0 uses AAC SBR for -q 0.3 and below. For -q 0.31 and above it uses AAC LC by default.

[singaiya]

A killer sample then ! Hope that the developers will take it in consideration. Of course, q 0.35 is not always transparent but to me, for portable use, it's more than enough. Thank you singaiya for posting your tests.

P.S. Maybe you should post your sample. See if /mnt's golden ears or anyone else could detect this problem at upper bitrates.

I still don't consider it a killer sample in the sense that other killer samples usually have sparse instrumentation to minimize masking effects and let artifacts stand out more. Here I'm not hearing specific artifacts except a general sound quality issue and perhaps less air and a more smeary dull sound. I kind of think that if I chose to spend enough time I could ABX every song on this album. This one just happens to be the first track on it and I gave it a go. But there's also the fact that I failed to ABX with the older encoder at this same bitrate.

I'll upload the first 30 seconds in the uploads forum.

[IgorC]

Also take in count that for same -q value 1.3.3 produces higher bitrate (approx. 5% here for complete discographies of bands). So if somebody compare old and new versions first step will be find equivalent -q values for each one.

[singaiya]

Yes, that is what melomaniac pointed out. Although in my case, I was coming from two versions ago (February 2007 encoder) and for the same -q value the bitrate went way down. That's why I retested using his -q value.

To summarize my results on this song:

Feb 2007 -q 0.25 -LC -> 103 kbps -> failed ABX
1.3.3.0 -q 0.25 -LC -> 65 kbps -> passed ABX (of course)
1.3.3.0 -q 0.35 -> 103 kbps -> passed ABX

[Lear]

I made a quick test using 1.3.3.0 on a couple of killer samples, and for one, applaud, it was easy to hear a difference. There's some noise added, that can be heard pretty much all the time. I used the following options for neroAacEnc in Foobar: "-q 0.4 -ignorelength -if - -of %d". The bitrate was 152 kbps.

CODE

foo_abx 1.3.3 report
foobar2000 v0.9.5.5
2008/10/21 20:01:53

File A: E:\Test\Test cases\applaud.wav
File B: E:\Temp\Foobar\Nero Q 0.40\applaud.mp4

20:01:53 : Test started.
20:02:37 : 01/01  50.0%
20:03:07 : 02/02  25.0%
20:03:22 : 03/03  12.5%
20:03:41 : 04/04  6.3%
20:03:59 : 05/05  3.1%
20:04:12 : 06/06  1.6%
20:04:26 : 07/07  0.8%
20:04:42 : 08/08  0.4%
20:04:49 : Test finished.

----------
Total: 8/8 (0.4%)

None of the other killer samples tested (castanets, fatboy, goldc, velvet) had the extra noise. Also made a very quick test with version 1.1.34.2 of the encoder, and it seemed to have the same problem (only tested applaud, and didn't ABX).

[/mnt]

I have seem to have a few possible ringing issues at 0.50 and even 0.55 with some tracks i tested out. Which seems to appear on cymbals.

Here is a few ABX test i did with the latest version of Nero AAC at q 0.50. Also these problems still appear on the older Feb 07 version aswell.

CODE

foo_abx 1.3.3 report
foobar2000 v0.9.6 beta 1
2008/10/21 20:17:57

File A: C:\Rips\Marilyn Manson - Lest We Forget (The Best Of)\01. The Love Song.flac
File B: C:\Temp\The Love Song q 0.50.mp4

20:17:57 : Test started.
20:18:14 : 01/01 50.0%
20:18:18 : 02/02 25.0%
20:18:22 : 03/03 12.5%
20:18:32 : 04/04 6.3%
20:18:41 : 05/05 3.1%
20:18:49 : 06/06 1.6%
20:18:56 : 07/07 0.8%
20:19:02 : 08/08 0.4%
20:19:11 : 09/09 0.2%
20:19:21 : 10/10 0.1%
20:19:25 : 11/11 0.0%
20:19:31 : 12/12 0.0%
20:19:32 : Test finished.

----------
Total: 12/12 (0.0%)

At the start of the track, the drum cymbals sound really annoying and very easy to notice; it could be ringing. Am not sure though, if its a ringing artifact since am not sensitive to ringing.

CODE

foo_abx 1.3.3 report
foobar2000 v0.9.6 beta 1
2008/10/21 20:21:07

File A: C:\Rips\Metallica - Ride The Lightning\07. Creeping Death.flac
File B: C:\Temp\Creeping Death q 0.50.mp4

20:21:07 : Test started.
20:21:41 : 01/01 50.0%
20:21:48 : 02/02 25.0%
20:21:52 : 03/03 12.5%
20:21:58 : 04/04 6.3%
20:22:03 : 05/05 3.1%
20:22:09 : 06/06 1.6%
20:22:15 : 07/07 0.8%
20:22:20 : 08/08 0.4%
20:22:28 : 09/09 0.2%
20:22:33 : 10/10 0.1%
20:22:41 : 11/11 0.0%
20:22:47 : 12/12 0.0%
20:22:48 : Test finished.

----------
Total: 12/12 (0.0%)

Same issue as above at around 0:10.

Anyway i will upload the samples ASAP, if anyone else wants to see if they can spot the problem aswell.

I have uploaded the samples

This post has been edited by /mnt: Oct 21 2008, 21:08

[muaddib]

Thank you very much for reporting issues. We'll try to address these and other quality issues in our next release.

I made a quick test using 1.3.3.0 on a couple of killer samples, and for one, applaud, it was easy to hear a difference.

Can you please give the link to the place where you obtained this sample?

[Lear]

Can you please give the link to the place where you obtained this sample?

Hm, that was years ago, but I'm pretty sure it is the same as applaud.wv here:

http://lame.sourceforge.net/download/samples/

[Supacon]

Solved compatibility issues with some hardware devices

Ironically, my new AAC files no longer will play in my car stereo or in iPods. I didn't have this problem before with nero's encoder. What's up with that? Am I alone in discovering this?

I'm using foobar2000 v9.5.2 with the builtin Nero AAC Encoder presets, quality of 0.31 and 0.4 and I've even tried forcing LC mode. No difference. Previously anything over 0.31 would playback fine in such hardware devices (anything over 0.31 tends to be encoded as AAC-LC).

[Sartre]

The latest version appears to apply a 18k (or lower) lowpass on everything, regardless of bitrate. I convert to 400kbps VBR with Foobar and never had the top end chopped off at this bit rate with previous versions.

Here's a spectral graph of the Pixies "Distance equals rate times time" converted from a FLAC file.

Original FLAC



Feb 2007 Neroaacenc 400kbps VBR



Sept 2008 Neroaacenc 400kbps VBR

[menno]

Can you post the exact command line you used?

[Sartre]

Can you post the exact command line you used?

I use Foobar which has a graphical interface to the program but selecting custom shows command line options for the previous setting which are (for Q1.00/~400kbps):

-q 1.00 -ignorelength -if - -of %d

The only file that didn't show a 18k lowpass was a 24bit/96khz wav file I also converted at Q 1.00 but the final bit rate was ~850kbps.

Edit: I just did a conversion of Pink Floyd - Money (from WAV) using the following command line in Foobar:

-br 400000 -ignorelength -if - -of %d

with the same result - a clear 18k lowpass.

This post has been edited by Sartre: Oct 30 2008, 16:35

[matt_t]

Thanks for this release, I have been waiting for it ever since I found out that Creative Zen and Xbox 360 compatibility issues would be fixed.

Can someone confirm that the Zen compatibility thing has indeed been fixed. I'm thinking about buying a new player, probably a Zen or a Sony Walkman, and this would be a useful piece of info.

Thanks!