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: mp3 or aac, at transparent levels (Read 5703 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

mp3 or aac, at transparent levels

I've been lurking this forum for a long time, but I decided to post before I rip a lot of CDs (to lossless as well, but that's FLAC without question).

So after a lot of time spent doing ABX tests (didn't save the logs, but done through foobar2000, at least 12 trials each, plus too many tests with small increment in quality anyway), I've come to a conclusion that it's more or less transparent to me after 175 kbps or so. I figured V2.0 is a good compromise, leaving a little room, just in case, as well. (By the way, using LAME 3.98.4 through XLD in OS X.) I've also been trying out AAC (QuickTime & CoreAudio, through XLD as well), and it's been transparent to me at more or less the same level (True VBR, encoder quality max, -q about 90), which I figured anyway, since at such high bitrates, both perform very similarly (though, I must mention how impressed I was with AAC at lower bitrates).

So with listening both at home and in iPhone/iPod in mind, which format do you guys suggest I encode in? At least to my ears, the quality is indistinguishable and file sizes are very similar. Any inherent advantages or disadvantages in either formats? mp3 is slightly more supported, but at this point, at least with the devices I use, AAC is just as well supported.

Thanks everyone!

mp3 or aac, at transparent levels

Reply #1 –
Quote from: Hresources on
I've been lurking this forum for a long time, but I decided to post before I rip a lot of CDs (to lossless as well, but that's FLAC without question).
...
(though, I must mention how impressed I was with AAC at lower bitrates).

So with listening both at home and in iPhone/iPod in mind, which format do you guys suggest I encode in?
...
at least with the devices I use, AAC is just as well supported.


I'm happy ripping to ALAC and having iTunes dynamically convert to 128 Kbps AAC when I sync to my iPod. It's a lot simpler than maintaining a parallel FLAC and MP3 library, and with iTunes not supporting FLAC, you'll be importing MP3 into iTunes if you rip to FLAC. After the ripping, if you change metadata in iTunes, you have to propagate it to the FLAC originals, and vice versa if you edit the FLAC files. Since you were impressed with AAC at lower bitrates, I'd recommend checking again using iTunes 10 to convert some files to 128 Kbps AAC and comparing them to ALAC in foobar2000. Worked for me, and it greatly simplified what I was originally doing when I started using iTunes/iPod a few months ago for the first time ever.

mp3 or aac, at transparent levels

Reply #2 –
Quote from: tima on
I'm happy ripping to ALAC and having iTunes dynamically convert to 128 Kbps AAC when I sync to my iPod. It's a lot simpler than maintaining a parallel FLAC and MP3 library, and with iTunes not supporting FLAC, you'll be importing MP3 into iTunes if you rip to FLAC. After the ripping, if you change metadata in iTunes, you have to propagate it to the FLAC originals, and vice versa if you edit the FLAC files. Since you were impressed with AAC at lower bitrates, I'd recommend checking again using iTunes 10 to convert some files to 128 Kbps AAC and comparing them to ALAC in foobar2000. Worked for me, and it greatly simplified what I was originally doing when I started using iTunes/iPod a few months ago for the first time ever.

The thing is, I have the FLAC files on an external drive, with MP3s on the internal. Being a laptop and having a lot of other things, it's not really feasible for me to carry FLAC copies around as well. I tend to rip to FLAC, fetch the metadata off MusicBrainz, then convert the FLAC with all the metadata to whichever format. I rarely have to change the metadata, but I guess if I ever get a desktop, that sounds like an awesome idea.

As for 128 kbps AAC, they do sound great. It's just I tend to notice the artifacts on crash cymbals almost all the time when it's louder. Not the biggest issue, but it tends to bother me a bit knowing it. Life was simpler not knowing I guess.

mp3 or aac, at transparent levels

Reply #3 –
Hresources,

If you are young and/or have good ears, please upload some samples of AAC at 128K having trouble with crash cymbals. Would be very helpful to the rest of us for our own ABX testing. Thanks.

mp3 or aac, at transparent levels

Reply #4 –
I posted a sample of Hysteria over here.

For constrained VBR at 128 kbps vs. FLAC
Code: [Select]
foo_abx 1.3.4 report
foobar2000 v1.0
2011/01/10 15:11:04

File A: Z:\Users\...\Downloads\xld\08 - Hysteria cut (constrained VBR 128).m4a
File B: Z:\Users\...\Downloads\xld\08 - Hysteria cut.flac

15:11:04 : Test started.
15:11:42 : 01/01  50.0%
15:12:12 : 02/02  25.0%
15:12:22 : 03/03  12.5%
15:12:35 : 04/04  6.3%
15:12:47 : 05/05  3.1%
15:13:08 : 06/06  1.6%
15:13:17 : 07/07  0.8%
15:13:33 : 08/08  0.4%
15:13:52 : 08/09  2.0%
15:14:05 : 09/10  1.1%
15:14:24 : 10/11  0.6%
15:14:55 : 11/12  0.3%
15:14:59 : Test finished.

 ---------- 
Total: 11/12 (0.3%)


For true VBR at -q 60 vs. FLAC
Code: [Select]
foo_abx 1.3.4 report
foobar2000 v1.0
2011/01/10 15:15:52

File A: Z:\Users\...\Downloads\xld\08 - Hysteria cut (true VBR -q 60).m4a
File B: Z:\Users\...\Downloads\xld\08 - Hysteria cut.flac

15:15:52 : Test started.
15:16:27 : 01/01  50.0%
15:16:40 : 02/02  25.0%
15:16:50 : 03/03  12.5%
15:17:04 : 04/04  6.3%
15:17:16 : 05/05  3.1%
15:17:30 : 06/06  1.6%
15:17:46 : 07/07  0.8%
15:17:57 : 08/08  0.4%
15:18:07 : 09/09  0.2%
15:18:21 : 10/10  0.1%
15:18:27 : 11/11  0.0%
15:18:37 : 12/12  0.0%
15:18:38 : Test finished.

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


Other samples I'll post later when I have time.

mp3 or aac, at transparent levels

Reply #5 –
I cannot seem to ABX that sample with your Quick Time encodings, but it's easy to ABX with Nero AAC at the same bitrate:
Code: [Select]
foo_abx 1.3.4 report
foobar2000 v1.1.2 beta 1
2011/01/10 23:52:16

File A: C:\Users\--------\Downloads\08___Hysteria_cut.flac
File B: C:\Temp\08___Hysteria_cut.m4a

23:52:16 : Test started.
23:52:39 : 01/01  50.0%
23:52:50 : 02/02  25.0%
23:53:10 : 03/03  12.5%
23:53:28 : 04/04  6.3%
23:53:38 : 05/05  3.1%
23:54:01 : 06/06  1.6%
23:54:09 : 07/07  0.8%
23:54:15 : 08/08  0.4%
23:54:28 : 09/09  0.2%
23:54:44 : 10/10  0.1%
23:54:58 : 11/11  0.0%
23:55:21 : 12/12  0.0%
23:55:22 : Test finished.

 ---------- 
Total: 12/12 (0.0%)

Code: [Select]
foo_abx 1.3.4 report
foobar2000 v1.1.2 beta 1
2011/01/10 23:57:46

File A: C:\Users\--------\Downloads\08___Hysteria_cut__true_VBR__q_60_.m4a
File B: C:\Temp\08___Hysteria_cut.m4a

23:57:46 : Test started.
23:57:52 : 01/01  50.0%
23:57:55 : 02/02  25.0%
23:58:00 : 03/03  12.5%
23:58:06 : 04/04  6.3%
23:58:11 : 05/05  3.1%
23:58:15 : 06/06  1.6%
23:58:18 : 06/07  6.3%
23:58:21 : 07/08  3.5%
23:58:26 : 08/09  2.0%
23:58:30 : 09/10  1.1%
23:58:36 : 10/11  0.6%
23:58:40 : 11/12  0.3%
23:59:02 : 11/13  1.1%
23:59:09 : 12/14  0.6%
23:59:12 : 13/15  0.4%
23:59:21 : 14/16  0.2%
23:59:33 : 15/17  0.1%
23:59:38 : 16/18  0.1%
00:00:04 : 17/19  0.0%
00:00:13 : 18/20  0.0%
00:00:16 : Test finished.

 ---------- 
Total: 18/20 (0.0%)

Also possible to ABX with LAME 3.98.4 at V5, due to warbling:
Code: [Select]
foo_abx 1.3.4 report
foobar2000 v1.1.2 beta 1
2011/01/11 00:02:55

File A: C:\Users\--------\Downloads\08___Hysteria_cut.flac
File B: C:\Temp\08___Hysteria_cut.mp3

00:02:55 : Test started.
00:03:02 : 01/01  50.0%
00:03:08 : 02/02  25.0%
00:03:23 : 03/03  12.5%
00:03:28 : 04/04  6.3%
00:03:33 : 05/05  3.1%
00:03:41 : 06/06  1.6%
00:03:53 : 07/07  0.8%
00:04:01 : 08/08  0.4%
00:04:08 : 09/09  0.2%
00:04:16 : 10/10  0.1%
00:04:34 : Test finished.

 ---------- 
Total: 10/10 (0.1%)
"I never thought I'd see this much candy in one mission!"

mp3 or aac, at transparent levels

Reply #6 –
Quote from: /mnt on
it's easy to ABX with Nero AAC at the same bitrate

How is it the same bitrate when it is 61% of the size of the Quicktime samples?

mp3 or aac, at transparent levels

Reply #7 –
Well, it's easy ABXable QuickTime encodes  not because of artifcats itself but AAC files have less amplitude.
Replaygain results:
Original: -10.03 dB
true vbr q60: - 9.59 dB.

It's might be not artifacts at all but aplitude differences.
I suggest to redo the test previously lower amplitude of source in any audio editor. This way Apple encoder won't apply its filtering.

mp3 or aac, at transparent levels

Reply #8 –
Quote from: greynol on
Quote from: /mnt on
it's easy to ABX with Nero AAC at the same bitrate

How is it the same bitrate when it is 61% of the size of the Quicktime samples?


My sample is at 134kbps avg, maybe the transcoding software or QuickTime pads about 60kbps on the MP4 tag for embeded artwork.

"I never thought I'd see this much candy in one mission!"

mp3 or aac, at transparent levels

Reply #9 –
Quote from: greynol on
Quote from: /mnt on
it's easy to ABX with Nero AAC at the same bitrate

How is it the same bitrate when it is 61% of the size of the Quicktime samples?


Nero -q0.41 produces the same file size as Apple --tvbr 60 for metal music for big amount of files. ~130 kbps.

mp3 or aac, at transparent levels

Reply #10 –
I was going by the reported file size of 80.67K.  Has anyone tried to repack the QT files to eliminate the space being attributed at padding?

@IgorC below and /mnt for explaining, thanks.

mp3 or aac, at transparent levels

Reply #11 –
Yes, I've just repacked the tvbr 60 file and it has shrinked to 80 KByte. The sample is quite short 5 sec that's why padding took some size.
 

mp3 or aac, at transparent levels

Reply #12 –
I'll try ABXing with RG on later, but I'm just wondering if any of you had any thoughts on the original question... I'd be encoding it around 180-190 kbps anyway, when it's transparent to me regardless, so yeah. Thanks.