Should i start using AAC? Can you notice a better quality? Options

[RobertoDomenico]

It's not an actual fault of Lame, the issue is to do with streaming over 'Home Sharing". Fact is Apple won't or can't fix it so i would avoid Lame Mp3 if one was to be using Apple devices. In terms of sound quality AAC & Lame MP3 are going to be pretty much identical outside of ABX testing.

This post has been edited by RobertoDomenico: Mar 25 2012, 05:45

[aprofromindia]

The modern lossy bitrate is far too bloated. Comparisons for mp3 Vs AAC are difficult as the average bitrate is now 250 .. 300k . Several years back the motivation of AAC and others was to provide near transparency @ 128k and better handling for problem samples where mp3 192k wasn't cutting it. Other encoders used VBR to implement these improvements. An early example was the MPC encoder which gave great quality @ 160..200kbit

Things where really interesting back then on the lossy side of things - 2002 ~ 2006. The other thing is mp3 was always competitive at 192k if you don't count rare problem samples. Even then it could probably sound very acceptable and satisfy 90 something % of people . This is probably true even for the old CBR 192 encodings.

Where its at today you could have just stuck it out with 256 CBR mp3 ten yrs ago , ignored any audio lossy development since and still be competitive .

To me QT AAC (-V 91 -q 2) sounds more compressed than 3.99.5 LAME (-V 2, -M j -q 0); my ABX between wav and qt aac below : -

foo_abx 1.3.4 report
foobar2000 v1.2.6
2013/06/14 20:47:35

File A: C:\Users\Apro\Desktop\3TalkAboutLove.wav
File B: C:\Users\Apro\Desktop\3TalkAboutLove.mp4

20:47:35 : Test started.
20:48:09 : 01/01 50.0%
20:48:53 : 02/02 25.0%
20:49:25 : 03/03 12.5%
20:50:06 : 03/04 31.3%
20:50:43 : 04/05 18.8%
20:51:16 : 05/06 10.9%
20:51:33 : 06/07 6.3%
20:53:03 : 06/08 14.5%
20:53:55 : 07/09 9.0%
20:55:33 : 08/10 5.5%
20:55:58 : 09/11 3.3%
20:56:12 : 10/12 1.9%
20:56:29 : 11/13 1.1%
21:05:47 : Test finished.

----------
Total: 11/13 (1.1%)

[TomasPin]

To me QT AAC (-V 91 -q 2) sounds more compressed than 3.99.5 LAME (-V 2, -M j -q 0)

You should remove the "-M j" and "-q 0" from the command line, as the defaults in LAME are tuned for best performance. Besides, the -q setting is irrelevant for VBR encoding. So, just use "V2".

That being said, either something's wrong when decoding AAC on your equipment and that's the difference you're hearing, or you have pretty privileged hearing. At that quality the result should be transparent in most situations. Perhaps you could post the sample you used so the rest of us can test as well. Make sure it's <= 30 seconds long.

[greynol]

It doesn't matter if -mj and -q0 are there. They do absolutely nothing when using VBR.

[TomasPin]

It doesn't matter if -mj and -q0 are there. They do absolutely nothing when using VBR.

So why include them?

Edit: I know I said "should remove" when in fact there's no problem in leaving them there. So, point taken.

This post has been edited by TomasPin: Jun 15 2013, 21:03

[db1989]

It should be specified that “-M” is not a valid switch. So, not only would the actual switch be pointless to include (or yes, technically, remove) here, but in fact, the incorrect version used by aprofromindia would never do anything because it’s not a recognised option.