Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Nero LC-AAC vs. aoTuV beta 5 (Read 16419 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Nero LC-AAC vs. aoTuV beta 5

Hello everyone!

I just conducted a tiny listening test to decide which bitrates to choose for my portable player (Trekstor i.Beat organix - MP3, WMA, Vorbis) and the mobile phone (Sony Ericsson K610i, MP3, WMA, HE-AACv2) in the future. Since both devices are equipped with small 512 MB flash storages I keep encoding to low, not implicitly transparent bitrates, to be able to squeeze satisfying amounts of music onto them. Up to this point the bitrates and codecs of choice were 80 kbps aoTuV beta 5 for the portable and 64 kbps HE-AAC for the mobile, with album ReplayGain irreversibly applied to the files themselves using foobar's converter. These encodings are gonna be replaced by track ReplayGain and LC-AAC instead of the HE solution, due to the mobile heavily draining batteries during playback. The only question was whether to encode to 64 or 80 kbps this time, hence I ABXed an extremely well-mastered and highly dynamic song produced by Rush, "Turn the Page" of the "Hold Your Fire" album.

Please note that this test was neither about transparency nor about representative results about the encoders in general, they were just for comparison whether there are clearly to distinguish differences between the two different quality settings, differences that would make encoding to 80 instead of 64 kbps worthwhile in conjunction with a flash-based portable player and its common headphones/earbuds. Talking about earbuds, the test was performed with average €30 headphones (Sennheiser HD 457) covering my two spoons.

A very noticeable and therefore good to ABX moment in "Turn the Page" are the 5th and 6th second of this small sample, the bang of the drums heard there sounds dull on extremely low bitrate encodings, hence I kept using it for all 5 tests I conducted so far. The results were quite interesting:

aoTuV beta 5 -q0:
Code: [Select]
foo_abx 1.3 report
foobar2000 v0.9.4.2
2007/03/28 15:32:42

File A: D:\Dokumente und Einstellungen\Tobias\Desktop\08 - Turn the Page.ogg
File B: S:\FLAC\Rush\Hold Your Fire\08 - Turn the Page.flac

15:32:42 : Test started.
15:33:05 : 01/01  50.0%
15:33:21 : 02/02  25.0%
15:33:57 : 03/03  12.5%
15:34:23 : 04/04  6.3%
15:34:59 : 05/05  3.1%
15:36:21 : 06/06  1.6%
15:37:00 : 07/07  0.8%
15:37:05 : Test finished.

 ---------- 
Total: 7/7 (0.8%)
Nero -lc -q 0.15:
Code: [Select]
foo_abx 1.3 report
foobar2000 v0.9.4.2
2007/03/28 15:38:38

File A: D:\Dokumente und Einstellungen\Tobias\Desktop\08 - Turn the Page.m4a
File B: S:\FLAC\Rush\Hold Your Fire\08 - Turn the Page.flac

15:38:38 : Test started.
15:40:24 : 01/01  50.0%
15:41:13 : 02/02  25.0%
15:42:28 : 03/03  12.5%
15:42:45 : 04/04  6.3%
15:43:54 : 05/05  3.1%
15:44:28 : 06/06  1.6%
15:46:00 : 07/07  0.8%
15:46:03 : Test finished.

 ---------- 
Total: 7/7 (0.8%)
aoTuV beta 5 -q1:
Code: [Select]
foo_abx 1.3 report
foobar2000 v0.9.4.2
2007/03/28 16:01:26

File A: D:\Dokumente und Einstellungen\Tobias\Desktop\08 - Turn the Page80.ogg
File B: S:\FLAC\Rush\Hold Your Fire\08 - Turn the Page.flac

16:01:26 : Test started.
16:02:01 : 01/01  50.0%
16:02:39 : 01/02  75.0%
16:05:06 : 01/03  87.5%
16:05:09 : Test finished.

 ---------- 
Total: 1/3 (87.5%)
Nero -lc -q 0.18:
Code: [Select]
foo_abx 1.3 report
foobar2000 v0.9.4.2
2007/03/28 15:56:16

File A: D:\Dokumente und Einstellungen\Tobias\Desktop\08 - Turn the Page80.m4a
File B: S:\FLAC\Rush\Hold Your Fire\08 - Turn the Page.flac

15:56:16 : Test started.
15:57:19 : 01/01  50.0%
15:57:58 : 01/02  75.0%
15:58:36 : 01/03  87.5%
16:00:22 : 01/04  93.8%
16:00:29 : Test finished.

 ---------- 
Total: 1/4 (93.8%)
The two 64 kbps encodings against each other:
Code: [Select]
foo_abx 1.3 report
foobar2000 v0.9.4.2
2007/03/28 16:16:10

File A: D:\Dokumente und Einstellungen\Tobias\Desktop\08 - Turn the Page.m4a
File B: D:\Dokumente und Einstellungen\Tobias\Desktop\08 - Turn the Page.ogg

16:16:10 : Test started.
16:19:59 : 01/01  50.0%
16:21:10 : 01/02  75.0%
16:23:08 : 01/03  87.5%
16:23:12 : Test finished.

 ---------- 
Total: 1/3 (87.5%)

The results reveal drastic quality improvements of both encoders if average bitrates of ~64 kbps are increased to ~80 kbps. While I didn't have any problems ABXing the dull sound of the drums at 64 kbps with both encoders, I was entirely unable to find clear differences between lossy @ 80 kbps and FLAC. Even more interesting, aoTuV Vorbis and Nero LC-AAC turned out being equal competitors if released on this sample, as you can see in the last test.

The conclusion is quite obvious: 80 kbps are definitely the better choice if a good compromise between file sizes and quality is needed, at least in situations like these where higher bitrates aren't an option at all. This, of course, only applies to the non-audiophile user (like me) who hasn't trained his hearing for ABXing artifacts at low bitrates, and if the files are gonna be played back by typical portable devices instead of high-quality equipment. If all these points are valid, the one or other user might find my results helpful, hence I posted them here.

Nero LC-AAC vs. aoTuV beta 5

Reply #1 –
1st, Nero AAC is not supposed to use LC-AAC at 64kbps. Next, as far as I tested, Nero -q0.325 produces nearly 80kbps LC-AAC which is equivalent of aoTuV -q1. VBR encoder should be compared with (almost) same quality setting.

One big con of low bitrate aoTuV is imperfect stereo. This might be inevitable. And I can easily find another problems on this sample and orion_ii.

Nero LC-AAC vs. aoTuV beta 5

Reply #2 –
Quote from: haregoo on
1st, Nero AAC is not supposed to use LC-AAC at 64kbps. Next, as far as I tested, Nero -q0.325 produces nearly 80kbps LC-AAC which is equivalent of aoTuV -q1. VBR encoder should be compared with (almost) same quality setting.

You might have missed the -lc switch, which enforces encoding to LC-AAC:
Quote
Nero -lc -q 0.15:
Nero -lc -q 0.18:

The -lc settings are independent of the default quality ones. -lc -q 0.15 averaged around 65 kbps, while -lc -q 0.18 went for ~76 kbps during my tests. As mentioned above, SBR's power hunger seriously sucks the limited battery life of the mobile phone, forcing me to recharge the accumulator noticeably often.

Nero LC-AAC vs. aoTuV beta 5

Reply #3 –
Quote from: Junon on
The -lc settings are independent of the default quality ones. -lc -q 0.15 averaged around 65 kbps, while -lc -q 0.18 went for ~76 kbps during my tests. As mentioned above, SBR's power hunger seriously sucks the limited battery life of the mobile phone, forcing me to recharge the accumulator noticeably often.

Additional -lc switch isn't recommended. Nero chooses the best method at given bitrate. That means your setting isn't the best setting of Nero.

Quote
Nero -lc -q 0.18:

Average bitrate of "-lc -q 0.18" is not 80kbps. You can't call this 80k while you call aoTuV q1 80k. Try your settins with other samples. That's far from 80k.

Nero LC-AAC vs. aoTuV beta 5

Reply #4 –
Quote from: haregoo on
Additional -lc switch isn't recommended. Nero chooses the best method at given bitrate. That means your setting isn't the best setting of Nero.
Nero's default setting, which is HE-AAC at 64 kbps, is of no use in this certain case, as mentioned above when I was talking about SBR being a problem in conjunction with the SonyEricsson thing. Hence the -lc switch had to be used in order to encode to the 64 kbps AAC profile needed for the test. At 80 kbps there's no difference at all since both the default and my custom settings use the same profile anyway. The only thing that needs adjustment is the command line.

Quote
Average bitrate of "-lc -q 0.18" is not 80kbps. You can't call this 80k while you call aoTuV q1 80k. Try your settins with other samples. That's far from 80k.
The Nero encoder's average bitrate drastically varies depending on the contents it progresses, way more than aoTuV, hence it's difficult to draw a general conclusion with which setting a certain file size is reached. -lc -q 0.18 usually stays slightly below 80 kbps, while 0.19 is often above 85 kbps. The music you commonly listen to might not lead to results comparable to mine, which mainly consists of Hard Rock and Metal. Especially the latter is notorious for making the bitrates explode.

Some examples, encoded with the -lc 0.18 switch:

Code: [Select]
[b]The test sample, Rush's "Turn the Page":

[/b]0:10.121 (446348 samples)
44100 Hz
2
79 kbps
AAC
AAC LC
lossy
Nero AAC codec / Feb 12 2007

[b]Audioslave - Revelations (whole album):

[/b]48:28.960 (128285136 samples)
44100 Hz
2
76 kbps
AAC
AAC LC
lossy
Nero AAC codec / Feb 12 2007

[b]Aerosmith - Pump (whole album):

[/b]47:43.773 (126292404 samples)
44100 Hz
2
80 kbps
AAC
AAC LC
lossy
Nero AAC codec / Feb 12 2007

[b]Beto Vazquez Infinity - Infinity (whole album). [/b]This album supports your opinion with its comparably low average bitrate:

50:12.867 (132867420 samples)
44100 Hz
2
73 kbps
AAC
AAC LC
lossy
Nero AAC codec / Feb 12 2007

Nero LC-AAC vs. aoTuV beta 5

Reply #5 –
AFAIK SE k610 have full support for HE-AAC and PS

Nero LC-AAC vs. aoTuV beta 5

Reply #6 –
Quote from: starcy on
AFAIK SE k610 have full support for HE-AAC and PS

Please don't miss my reasons for using the -lc profile:
Quote
These encodings are gonna be replaced by track  ReplayGain and LC-AAC instead of the HE solution, due to the mobile  heavily draining batteries during playback.[...]As mentioned above, SBR's power hunger seriously sucks the limited  battery life of the mobile phone, forcing me to recharge the  accumulator noticeably often.

The SBR-HQ    (High Quality) implementation used by the phone needs a lot more processing power than SBR-LP (Low Power) or even plain LC-AAC. While this isn't a problem for PC playback, it turns out being one if you often and ardously listen to music via your mobile. Some additional informations about this matter: https://datatype.helixcommunity.org/2005/aacfixptdec and http://www.chiariglione.org/mpeg/technolog...4-sbr/index.htm

Nero LC-AAC vs. aoTuV beta 5

Reply #7 –
Quote from: Junon on
Nero's default setting, which is HE-AAC at 64 kbps, is of no use in this certain case, as mentioned above when I was talking about SBR being a problem in conjunction with the SonyEricsson thing. Hence the -lc switch had to be used in order to encode to the 64 kbps AAC profile needed for the test.

I knew it. I just say that's not best encoding.

Quote
The Nero encoder's average bitrate drastically varies depending on the contents it progresses, way more than aoTuV, hence it's difficult to draw a general conclusion with which setting a certain file size is reached. -lc -q 0.18 usually stays slightly below 80 kbps, while 0.19 is often above 85 kbps.

True. There is no nominal bitrate for Nero. But "-lc -q 0.18" is definitely not enough for 80kbps. I've searched frequently used settings of Nero when it's out. I use several public listening test samples to calculate bitrate, so my favorites music doesn't matter.

Nero LC-AAC vs. aoTuV beta 5

Reply #8 –
Hi Junon,

While I agree with your using the -lc switch to produce 64kbps LC files (the only way to go with Nero), I think you should try using the command line "-q 0.31" as well as your own"-q 0.18 -lc" at 80kbps and see which is better.  In my experience, all files encoded at -q 0.31 have turned out as LC profile so there's no need to use the switch to force it, and the difference in the q value suggests there may be audible improvement.  That was my impression in sighted listening (which counts for nothing at HA), but it's still worth a try if you have the time and motivation.

Haregoo, I wouldn't be surprised if "-q 0.18 -lc" is producing 80kbps files.  When you use the -lc switch, the bitrate goes up significantly for a given q value, or the q value goes down a lot for a given bitrate.

Stephen

Nero LC-AAC vs. aoTuV beta 5

Reply #9 –
I'll certainly give your suggestion a shot, especially since some more extensive testing with a wide variation of different music genres (Classic, Impressionism, Pop, Crossover, lots of Rock and Metal subgenres and game/movie soundtracks) revealed that -lc 0.18 does tend to average a bit below the 80 kbps mark, in this case at 74 kbps. Hence, for comparison I'll conduct a few more encodings using your -q 0.31 setting.

Nero LC-AAC vs. aoTuV beta 5

Reply #10 –
Quote from: steveh on
Hi Junon,

While I agree with your using the -lc switch to produce 64kbps LC files (the only way to go with Nero), I think you should try using the command line "-q 0.31" as well as your own"-q 0.18 -lc" at 80kbps and see which is better.  In my experience, all files encoded at -q 0.31 have turned out as LC profile so there's no need to use the switch to force it, and the difference in the q value suggests there may be audible improvement.  That was my impression in sighted listening (which counts for nothing at HA), but it's still worth a try if you have the time and motivation.

Haregoo, I wouldn't be surprised if "-q 0.18 -lc" is producing 80kbps files.  When you use the -lc switch, the bitrate goes up significantly for a given q value, or the q value goes down a lot for a given bitrate.

Stephen

This is all nonsense. If NeroEnc choses the LC profile in default mode (q > 0.3), the quality of the output is *equal* to the quality when using -lc with a q that gives the same bitrate. Nero simply chose to apply a different scale for q in default mode (dyamic profile) than for the forced LC profile. You can probably find a mathematical formula to translate from the default q scale to the forced LC one if you really want to.

Nero LC-AAC vs. aoTuV beta 5

Reply #11 –
-lc -q 0.18 and -q 0.31 deliver almost if not completely identical results. Trying to ABX them vs. each other is pointless, even the average bitrates of the 3 albums I've tested up to this point were the same (rounded to natural numbers) with both settings. But nonetheless, thanks for your suggestion, it made me aware of the fact that using the -lc switch isn't necessary at the desired bitrate range.

Nero LC-AAC vs. aoTuV beta 5

Reply #12 –
Quote from: Junon on
-lc -q 0.18 and -q 0.31 deliver almost if not completely identical results. Trying to ABX them vs. each other is pointless


Ok, fair enough.  It was only idle speculation on my part, but I'm glad you tried it out.  I guess I just don't like the idea that different quality values result in the same audible quality, or the same quality value resulting in two different audible qualities.
 

Nero LC-AAC vs. aoTuV beta 5

Reply #13 –
It's been discussed at length a year back when the Nero AAC encoder came out. Basically, it's this: you can have two scales, a quality scale and a bitrate scale (quality here meaning an internal measure, a target for the encoder to reach, not necessarily audible audio quality). With the -lc or -he switches (i.e. without changing the coding technique) you get a quality scale which is also a bitrate scale: higher quality=higher bitrate.

But with the combined scale (letting Nero pick the encoder), there comes a point where you switch from HE to LC (this is around 80kbps HE, or q=0.31 in the combined method).

If you would now continue with the same quality in LC, you'd need much higher bitrates - something like 130kbps. People would complain, "why can't I encode at 100 kbps?". So instead the decision was made to get a smooth bitrate transition, which means jumping back in quality some bit when going to LC - so q=0.31 in the combined measure "maps" to q=0.18 or so in the LC-model.

In the above-mentioned discussion, Garf provided a link to a graph that showed this.