High Bitrate AAC / LAME MP3 ABX Tests

2021-05-30 12:08:46

With my new audio setup (AES directly into some Genelec 8341 AP, Genelec 7370 APM sub, partially treated room) I thought it would be interesting to do some high bitrate tests to check transparency.

I've pick a single track, but a very popular "normal" music track. Jamiroquai "Virtual Insanity", which I hate and now hate even more.

I found the AAC very difficult to tell apart, but according the the ABX the MP3 V0 was closer. When I get more time I'll test more tracks.

Track was ripped from the initial pressing of the CD then converted to the following:

Code: [Select]

Duration : 5:40.933 (15 035 160 samples)
Sample rate : 44100 Hz
Channels : 2
Bitrate : 259 kbps
Codec : MP3
Codec profile : MP3 VBR V0
Encoding : lossy
Tool : LAME3.100
Tag type : id3v2.3|id3v1
Embedded cuesheet : no

and

Code: [Select]

Duration : 5:40.933 (15 035 160 samples)
Sample rate : 44100 Hz
Channels : 2
Bitrate : 263 kbps
Codec : AAC
Codec profile : AAC LC
Encoding : lossy
Tool : Music 1.0.6.10
Apple SoundCheck gain : -0.66 dB

Results:

Code: [Select]

foo_abx 2.0.6d report
foobar2000 v1.6.6 beta 5
2021-05-30 11:49:48

File A: 1.01. Jamiroquai - Virtual Insanity.m4a
SHA1: 019d6df3d9170b9bab376f63ea39b23a17a222ab
File B: 01 Virtual Insanity.m4a
SHA1: bce99dc07d6dcee106c4f27cad2fa6e82b3a886a

Output:
Default : AES / Genelec
Crossfading: NO

11:49:48 : Test started.
11:51:12 : 01/01
11:51:41 : 02/02
11:52:09 : 03/03
11:52:49 : 03/04
11:53:06 : 04/05
11:53:47 : 05/06
11:54:34 : 06/07
11:55:31 : 07/08
11:56:11 : 08/09
11:56:38 : 09/10
11:56:54 : 10/11
11:57:10 : 11/12
11:57:23 : 12/13
11:57:42 : 13/14
11:57:57 : 14/15
11:58:16 : 15/16
11:58:16 : Test finished.

 ---------- 
Total: 15/16
p-value: 0.0003 (0.03%)

 -- signature -- 
07ef7306d81ff01d0033468840e51617cf0a8797

Code: [Select]

foo_abx 2.0.6d report
foobar2000 v1.6.6 beta 5
2021-05-30 11:31:48

File A: 1.01. Jamiroquai - Virtual Insanity.m4a
SHA1: 019d6df3d9170b9bab376f63ea39b23a17a222ab
File B: 1.01. Jamiroquai - Virtual Insanity.mp3
SHA1: 2bcca0e759477d70ffa16c12fa9e7105c2487eb0

Output:
Default : AES / Genelec
Crossfading: NO

11:31:48 : Test started.
11:32:46 : 01/01
11:33:13 : 02/02
11:33:43 : 03/03
11:34:05 : 04/04
11:34:22 : 05/05
11:34:42 : 06/06
11:35:03 : 07/07
11:35:29 : 08/08
11:35:43 : 09/09
11:36:06 : 10/10
11:36:29 : 11/11
11:36:48 : 12/12
11:37:00 : 12/13
11:37:18 : 13/14
11:37:39 : 14/15
11:37:58 : 14/16
11:37:58 : Test finished.

 ---------- 
Total: 14/16
p-value: 0.0021 (0.21%)

 -- signature -- 
2c387dd176862a31d1fa5bee348bf41bece55341

Re: High Bitrate AAC / LAME MP3 ABX Tests

Reply #1 – 2021-05-30 13:14:35

Thanks for sharing the experience

Quote from: NateHigs on 2021-05-30 12:08:46

I found the AAC very difficult to tell apart, but according the the ABX the MP3 V0 was closer. When I get more time I'll test more tracks.

ABX doesn't tell that. 14/16 vs 15/16 doesn't necessary imply that the latter is worse. It's only a probability of guessing, not a score quality. If you want to rank lossy files, you should use an ABC/HR software rather than a simple ABX tool.

Re: High Bitrate AAC / LAME MP3 ABX Tests

Reply #2 – 2021-05-31 22:06:19

Quote from: guruboolez on 2021-05-30 13:14:35

Thanks for sharing the experience

Quote from: NateHigs on 2021-05-30 12:08:46
I found the AAC very difficult to tell apart, but according the the ABX the MP3 V0 was closer. When I get more time I'll test more tracks.
ABX doesn't tell that. 14/16 vs 15/16 doesn't necessary imply that the latter is worse. It's only a probability of guessing, not a score quality. If you want to rank lossy files, you should use an ABC/HR software rather than a simple ABX tool.

I know, it was only a bit of a light hearted quip 👍

Certainly I think the AAC sounded more pleasing to me - I think I may actually have pressed to wrong button on some of the MP3 tests 😂 I’ll do another track tomorrow

Notice