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: [TOS #8] MP3 sounds better than AAC (Read 12729 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

[TOS #8] MP3 sounds better than AAC

Hey guys, I just did some extensive ABXing on "ABXer", and came out with some interesting results that i'd like to share and see what people have to say. First of all, I got into this because I am going to bee doing some ripping and encoding shortly and decided to really investigate file formats and bit rates in depth. So I did a lot of research and eventually got ABXer which ultimately gave me results I did not expect.

I didn't go into the testing with much confidence, I guessed that I would be able to tell the different between anything above 128 kbps and lossless. I compared 64 kbps, 128 kbps, 192, kbps, 320 kbps, in both AAC and MP3 as well as the original file that I used to get these compressed versions which was Apple Lossless. Before this testing I semi decided that I would rip my future stuff with AAC since is apple's main lossy format and its supposed to be better than MP3 at the same bit rate.

I ended up exceeding what I expected I was capable of and being able to distinguish between 192 kbps and 320 kbps, as well as 320 kbps and lossless (17 out of 20 for 320 vs lossless, is the only one I took screen shot of, earlier I did more tests that had some even better scores). When ABXing the higher bit rates and lossless, I found it very frustrating, but when I was able to focus and listen in exactly the right way, I sort of intuitively could tell the difference. It wasn't really a conscious thing, and the harder I tried the worse I got, but I eventually got the hang of it after a couple tests. I ended up doing it rather fast without thinking, using sort of "peripheral hearing", because if I focused in the wrong way and tried to listen for something that gives it away, I would get bad test results. The only way i can describe the difference would be that I perceived the higher bit rates usually as being louder, though i dont know if they actually where. I cannot say that lossless sounds better than 192 kbps, but I can tell the difference between them.

Anyways, what surprised me the most was when I tested the MP3s vs AACs of the same bit rate with each other. I used the ABXer to play them back without me knowing which was which, and I always chose the MP3 as better sounding. I didnt really understand why this was the case because i preferred it at 320 kbps and 64 kbps. Also, i found it easier to tell the difference between aac and mp3 files of the same bit rate than the same format at 320 vs 192. I found the MP3s to sound possibly livelier, punchier, louder, and at lower bitrates like 64 kbps I found it held the highs and lows better. Now, I thought maybe this could have something to do with the encoding software/codecs. But I used LAME for the MP3 and Apple Core i guess for the aac. The program I used was XLD. So I am wondering, can anybody explain why the MP3 sounded so much better to me?

Also, if anybody is interested the song I used for testing was Primal Directive by The Contortionist.

[TOS #8] MP3 sounds better than AAC

Reply #1 –
LAME MP3 sounded "better" than QuickTime AAC @ 64 kbps?

Something was very wrong with your AAC encoding...

Can you post a <30 second sample of that song in a lossless format (preferably FLAC, which XLD can convert to) to the Uploads forum?
"Not sure what the question is, but the answer is probably no."

[TOS #8] MP3 sounds better than AAC

Reply #2 –
I remember an experiment where a lot of younger people preferred mp3 over the original. The explanation then was that the listeners were accustomed to the softening of impulses which mp3 at very moderate bitrates does as a tendency, especially with encoders available earlier these years.

But at 64 kbps Lame resamples to 24 kHz for efficient encoding in order to suppress artifacts, and the 11 kHz lowpassed music is expected to be audible for the worse, as is the increased smearing of impulses.

Out of curiosity I just encoded a track I like (Jennifer Warnes: Lights of Louisianne) with Lame 3.98.4 using -b 64. I am impressed: sound is very good to me, I can't hear the 11 kHz lowpass. Though this song probably isn't very demanding I'm really impressed.
lame3995o -Q1.7 --lowpass 17

[TOS #8] MP3 sounds better than AAC

Reply #3 –
Wait...  So you ABXed 192kbps vs 320kbps?  You can't directly ABX two different lossy files.  Well, you can do that but it doesn't show anything other than you having the ability to distinguish between the two lossy files.  It doesn't mean that one is better than the other simply because you have no idea which is which.  So being able to ABX 192 and 320kbps does not automatically mean that 320kbps was better, it just means that you could hear the difference between the two and, without knowing which one was the higher bitrate track, it is impossible to tell which one you thought had better audio quality.  The only time that you can really do this (at least that I can think of) is when you are lossy-to-lossy transcoding files and you want to determine how well the transcoded resulting file performs in comparison to the source lossy file.  Other than that, ABXing a 192kbps mp3 file from a 320kbps AAC file doesn't really show much other than you could hear a difference between the two.  It can't tell you anything about the quality of either track.  You can assume that the AAC track was higher quality since it used a larger bitrate but you can't know for sure.  You only know that you could hear differences between the two.

That is why, when you are determining lossy encoding quality results for yourself, you compare lossy to lossless.  Then again, you may not have done that since your description seems a little off in that regards.  It is also generally a good idea to post your results (you can use the code box) for all of your ABX tests to show any trends with increasing bitrate along with changes in encoders/formats.  Lastly, it is another good idea to upload a 30 second lossless test clip of the song you were using while also fully listing each and every encoder you were using along with their appropriate settings.  Just saying Lame and Apple AAC encoder isn't enough.  Is that Lame 3.90.3, 3.96.1, 3.97b1, 3.97b2, 3.98.4, or 4.0a2?  What version of the Apple (QuickTime) AAC encoder was used?  What settings did you use?  Was this all at 64kbps, 128kbps, 192kbps, and 320kbps CBR, ABR, VBR, VBR_constrained, or true VBR?

[TOS #8] MP3 sounds better than AAC

Reply #4 –
As said, ABXing lossy vs. lossy isn’t very useful; ABX tests are designed to evaluate perceptual transparency, whereas all your tests can ever suggest is that you have a preference for one of the files, not that it’s more transparent (that being the only definition of “sounds better” with any objective utility).

Please provide useful information on the different settings, as requested by kornchild2002; and note that, even if the tests were all conducted properly, many users would prefer text logs rather than pictures or vague anecdotes.

[TOS #8] MP3 sounds better than AAC

Reply #5 –
For more information about double-blind testing which includes a ranking (to demonstrate preference):
http://wiki.hydrogenaudio.org/index.php?title=ABC/HR

Note that while the HR portion is optional, we run into the previously described problems without it.

[TOS #8] MP3 sounds better than AAC

Reply #6 –
He was simply pointing out that for comparing quality, you used the wrong tool. You use ABX to prove that you can hear a difference between A and B, not to show that you prefer the sound of A over B.

[TOS #8] MP3 sounds better than AAC

Reply #7 –
Quote from: pdq on
He was simply pointing out that for comparing quality, you used the wrong tool. You use ABX to prove that you can hear a difference between A and B, not to show that you prefer the sound of A over B.


I used ABXer to see if I can tell the difference and I simultaneously jotted down which side I thought sounded better A or B, than after the test was complete I checked that side A was "this file" and side B was "that file". I dont see why you cannot use ABXer for both purposes?



You can download the track (flac) here: link removed  EDIT: How do I cut 30 seconds out of a song so I can upload it? What app for this?

Also, the Mp3 files say "encoded with LAME3.98" both the AAC and MP3 "encoder quality" were set to max in XLD, I didn't really change anything else.

I'd be interested in hearing what people prefer at 64kbps... MP3 or AAC.

[TOS #8] MP3 sounds better than AAC

Reply #8 –
Quote from: db1989 on
As said, ABXing lossy vs. lossy isn’t very useful; ABX tests are designed to evaluate perceptual transparency, whereas all your tests can ever suggest is that you have a preference for one of the files, not that it’s more transparent (that being the only definition of “sounds better” with any objective utility).


Anyway the OP results show that at least one of that lossy encodings is not transparent to him (at the given bitrate).
... I live by long distance.

[TOS #8] MP3 sounds better than AAC

Reply #9 –
Quote from: devon8822 on
Quote from: pdq on
He was simply pointing out that for comparing quality, you used the wrong tool. You use ABX to prove that you can hear a difference between A and B, not to show that you prefer the sound of A over B.



I used ABXer to see if I can tell the difference and I simultaneously jotted down which side I thought sounded better A or B, than after the test was complete I checked that side A was "this file" and side B was "that file". I dont see why you cannot use ABXer for both purposes?


As greynol noted, without a reference, you don't know which sounds better since you don't know what the sound is supposed to sound like.  In your test you don't even know which of the two files was audibly different from the source, let alone which sounded better.  Take a look at the FAQ entry greynol linked.  He wasn't doing that for no reason, he linked it because its something you need to understand if you're going to try and compare audio files

[TOS #8] MP3 sounds better than AAC

Reply #10 –
Quote from: devon8822 on
So I am wondering, can anybody explain why the MP3 sounded so much better to me?

is this a joke? maybe at 64kbps you prefer the worse quality