For this discussion.

Click to view attachment
Click to view attachment
Click to view attachment
Click to view attachment

• blackwater sample



The lowpass betrayed Apple encoding (audible on cymbals, more pleasant and sharp with Real encoder). The first part of the sample was OK for me with both encoders. Therefore, both encoding are good in my opinion (and my listening system, based on AC'97 and poor line-out).


• waiting sample



Here, the quality difference is really consequent. Apple AAC@128 is seriously distorted, whereas Real AAC@192 is really less annoying.

Results for Waiting, Apple vs Real:

Sample A: waiting apple.wav
Sample B: waiting real.wav

1 of 1, p = 0.5
2 of 2, p = 0.25
3 of 3, p = 0.125
4 of 4, p = 0.062
5 of 5, p = 0.031
6 of 6, p = 0.015
6 of 7, p = 0.062
7 of 8, p = 0.035
8 of 9, p = 0.019
9 of 10, p = 0.01
10 of 11, p = 0.0050
11 of 12, p = 0.0030
12 of 13, p = 0.0010
13 of 14, p = <0.001
14 of 15, p = <0.001
15 of 16, p = <0.001
16 of 17, p = <0.001
17 of 18, p = <0.001
18 of 19, p = <0.001
19 of 20, p = <0.001

COMMENTS: I concentrated on the first 4 seconds of the sample, listening to the singer's voice. Apple's encoder seems to create a sort of metallic or echoic lisp on the singer's higher notes ("I've been...."). Real does not have this effect. Also, the lowpass was noticeable here (though I'll admit that my >16Khz hearing isn't something to write home about, in my old age of 25 )

Blackwater: I was not able to ABX the two, probably as a result of my poor HF hearing, though the significance was borderline at the .05 level (14/20, p=.057.)

A slightly OT comment is that actual songs I have purchased from iTunes have artifacts that are not found in encodings made "at home" with the same CD and settings. Perhaps that's another thing we need to consider...I can upload comparison samples if anybody is interested in what I mean.

Thanks guruboolez and Cygnus X1 for taking the time to try them out. Seems like my results are backed up by yours.

Note to mods: maybe we should take these comments to where the rest of the conversation is? (http://www.hydrogenaudio.org/forums/index.php?showtopic=24568&st=25)

PS. In the AAC test Real did as good as iTunes on waiting. So the result here (with more bits Real is better) should not be surprising.

thanks to echo for encoding the samples and the comparison, and to guruboolez and Cygnus X1 for taking the time. These are useful results and samples to have.

One problem is: are we sure that the encoders we've tested are the same than the ones used for each stores? I never heard or analysed any Real AAC purchased file, but the only AAC@128 from iTunes Music Store I had was totaly different from all QuickTime AAC@128 encoders ever released (full frequencies up to 16000 Hz, and from 16 to 20 Khz, additional informations but at low energy).

In my opinion, comparing ATRAC/AAC encoders bundled with public software is not the good way to compare the sound quality of different on-line stores.

Guruboolez, my experience has been twofold in terms of encoders: many of the songs I bought from iTunes (which isn't many, actually ) sound much worse than the files I made afterwards once I purchased the actual CD. Of course, I don't normally rip things at 128kbps (I still use --aps), but I did so to compare the purchased file to the one I had ripped. In almost every case, a) iTunes purchased AAC file had ringing and flanging artifacts NOT found in my home-ripped versions, and b) most of the file exhibited a sharp 15-16Khz lowpass. Even with my HF hearing (not very sesnsitive above 16Khz, cutoff around 17.5), the lowpass is noticable on some songs. (Edit: I used QT 6.5 w/ iTunes 4.2 for comparisons).

However I recently bought a file that had content to 20Khz and sounded fine (no obvious flaws). Unfortunately, I accidentally deleted it somehow and can't remember what it was. So, I'm not sure what's going on with the different encoders, but suffice to say, my experience has been that iTunes music is inferior to the same rip done at home from a CD. In fact, with one of the albums (George Benson's greatest hits), the files sounded like they had been transcoded from 112kbps MusicMatch files!

Agree. 10 or more samples from the actual online stores should be tested, if we want to find out anything about the online stores qualities.

Is there a chance that this difference in the iTunes store can only be attributed to different encoder versions? I guess Apple upgraded their encoders they use one the store together with the iTunes version and most people here seem to use the old iTunes 4.2 version of the encoder. Or maybe is it a totally different encoder? Although I don't understand why they would have two different versions of their encoder, one for their own use and one for the public. And I guess noone can rule out that the original source is not a CD but something else.

Maybe at least Karl can tell us about the encoder Real uses. Is it the exact same encoder found in Real player? And what about the original sources? Are they CDs or something else provided by the music labels?

This can't be done in a public test since you would need to break both DRM systems... Or maybe if you burn the fiels you bought and put them in FLAC then.

I have implemented a AAC MP4 decoder able to decode Apple's m4a files.. I found out that ITune always uses Sin Window shape.. It doesn't switch to KB Window at all.. I think there is likely to be some aliasing problem with ITune.. which I can hear very well in some song samples..

echo, is it possible to get the original sample in order to evaluate (real Vs orig) Vs (apple Vs Orig)?

Don't know if the comparison between 2 ABX scores makes sens though.

-ec

As I said in my original post (in the other thread, the discussion seems to have moved here now) you can find it in ff123's page.

And about comparing ABX scores, no it doesn't make much sence... unless you are able to ABX codec A and not able to ABX the codec B. In this case I think you can safely say codec B is better than codec A.

edit for clarification: I mean ABXing original vs codec A and original vs codec B of course.

OK, thanks!

I'll try this.