Ipod: AAC vs Mp3 (256k)
Oct 19, 2003 at 11:00 PM Thread Starter Post #1 of 25

iamdone

Headphoneus Supremus
Joined
Apr 10, 2003
Posts
2,080
Likes
10
I just did a test last night using AAC at 256 (iTunes for pc) and mp3 LAME (EAC high setting LAME 3.93.1). I didn't have my ety E4-S with me, so I used Senn. px200 and a Headsave Transit amp. I used the first 4 songs off Radiohead's Kid A for my sample.

The AAC has a slightly wider soundstage and creates a unique surround effect but the center sound doesn't sound right. It sounds like the left and right channel are being pulled apart just a bit. The vocals sounded a little in the background.

The mp3 sounded a little narrower but the vocal were more up front and it had a crisper sound to it. It sounded just bit harsh compared to AAC. Overall mp3 just sounded more enjoyable than AAC. AAC sounds like you added stereo seperator to it. This probably works great at lower bit rates but not a 256.

I also tested WAV files on the Ipod to use as the original. The three files sound almost the same when doing side-by-side testing but again, the problem can up with AAC. The center vocals are not as forward as WAV or mp3. If you are not using a headphone amp, I don't think you'll be able to tell the difference because soundstage placement is much harded to detect without one. The mp3s had just a little less (barely noticeable) dynamic range than the WAV but since the soundstage is the same, I prefer it to AAC. AAC seems to have almost the same dynamic range as the WAV. If they can correct the soundstage in future upgrades, then AAC would be the winner.

Also after listening to AAC a little more, I notice another problem. I causes the bobbles in the head feeling that I don't get from mp3s. It kind of makes you feel a little dizzy. I know crossfeed can fix this but you don't need this for mp3s.
 
Oct 19, 2003 at 11:10 PM Post #2 of 25
Could you give us the output settings of the MP3's? Unfortunately EAC is PC only so for those not on that platform it would be good to know what the command line is. Is it close to --alt-preset extreme?

Also I believe each format has its sonic advantages and disadvantages. I'm currently getting a few heckles over at Hydrogen Audio for saying so yesterday, but if there are differences (especially at the bitrates we're likely talking about), I'd be interested on your feedback with better phones.

Thanks for the comments.
 
Oct 19, 2003 at 11:20 PM Post #3 of 25
With EAC, you can choose high or low setting if you don't use the command line. When the conversion screen comes on, I see a setting of 2. With the lower setting, it's at 5. I don't know if extreme would bring that level to 1. I used the high setting for all my mp3s. I hope this helps.

Also, the width of the soundstage sounded identical use mp3 and WAVs. AAC sound a little different in that area too.
 
Oct 19, 2003 at 11:32 PM Post #4 of 25
Just want to verify then, you're choosing a specific bitrate for the LAME MP3's (as ABR or CBR) and not actual VBR? I guessing the number then is the quality setting and not one of the presets? Sorry to digress from the actual comparison, but I'm trying to understand the competitors.
 
Oct 19, 2003 at 11:41 PM Post #5 of 25
Yes, I am using a fixed bit rate of 256k and a high quality setting for the encoding. I also have it set on high quality for the ripping error rate. Both slow down the encoding process but make excellent mp3s.
 
Oct 20, 2003 at 2:02 AM Post #6 of 25
Aside from your comparisons here, have you tried "--alt-preset extreme" or "--alt-preset fast extreme" to get a likely overall lower average bitrate/filesize, but better quality? This would be a true VBR scheme based on the songs complexity instead of the one you're using that is actually ABR. Someone correct me if I am wrong. LAME doesn't have a VBR 256 kbps setting does it?
 
Oct 20, 2003 at 1:44 PM Post #7 of 25
Yes it does but this would not produce better quality, just smaller file space. Even at 320, I've heard VBR can effect the dynamic range.
 
Oct 20, 2003 at 2:17 PM Post #8 of 25
Wierd, but I don't have a VBR setting for anything higher than 192Kbps with EAC and LAME 3.91xx (I don't remember it for 3.93 either, but I could be mistaken). I use --preset extreme with my 192 VBRs, and the quality was very, very close even on poorly engineered sources. This was with my Plextor, Pioneer, and Sony drives. iTunes has the ability to do VBR at 320, but I would think that kinda defeats the purpose of VBR in the first place, since the average size would be close to 320 anyway, right?
 
Oct 20, 2003 at 3:19 PM Post #9 of 25
Quote:

Originally posted by clutch
Wierd, but I don't have a VBR setting for anything higher than 192Kbps with EAC and LAME 3.91xx (I don't remember it for 3.93 either, but I could be mistaken). I use --preset extreme with my 192 VBRs, and the quality was very, very close even on poorly engineered sources. This was with my Plextor, Pioneer, and Sony drives. iTunes has the ability to do VBR at 320, but I would think that kinda defeats the purpose of VBR in the first place, since the average size would be close to 320 anyway, right?


No, it would be best to use the higher bit rate with either CBR or VBR. With CRB it would just stay at 320 even at the silent parts. With VBR, it will use a max of 320 but will lower when not needed (silent and less dynamic parts). This why it saves file space but can never be better than constant bit rate but maybe about the same. I believe it takes even longer to encode using VBR.

The reason I encode at 256k is that I've read many studies that claim it is very hard to tell any difference above this level and there is not reason to go higher. I haven't tested 320 but I have 60 GB of music encoded at 256k so I think I'll keep it this way. Maybe I'll play around with AAC some more using some different encoders and see if I can get better results.

The reason for this test is I just thought that now that iTunes is out for Windows, it might be a good time to switch to AAC if it proved better.
 
Oct 20, 2003 at 4:33 PM Post #10 of 25
We may want to clear up the use of VBR here. VBR can mean anything not CBR or it can mean something separate from ABR (Average Bit Rate). I'm about 95% LAME doesn't have a true VBR scheme at bitrate settings. In fact it doesn't really make sense to have a bitrate setting for VBR. So LAME (and iTunes) uses an ABR setting at the various bitrates to maximize quality while still outputting a file the final same size as CBR. Oversimplified on ABR the same amount of bitrates/frames will be below the selected bitrate as above. This is why 5 minutes of music will take up the same space as 5 minutes of silence. But a true VBR setting like the presets in LAME may produced a better file because the resulting bitrates will not be limited by a base bitrate. On most music it's unlikely a true VBR scheme will be better than ABR 256 because you're so close to the max, 320, already. On a very complex piece you may be in 224-320 range the whole song. At least theoretically, a preset could have the entire piece at 320 if needed. And if not, it could be far below.

See Hydrogen's recommended settings. Getting back to the original post, this could mean you'd prefer MP3 even more.
 
Oct 20, 2003 at 4:56 PM Post #11 of 25
That might explain it. Are the AAC encoders that use CBR?
 
Oct 20, 2003 at 5:32 PM Post #12 of 25
There's some debate about iTunes and QT encoding. AAC is suppose to be VBR be nature, but iTunes/AAC isn't as defined separately from ABR. Someone played back a few tracks on WinAmp using a AAC plugin recently and said the bitrate fluctuated, so I'm guessing it's ABR. Others though have called it CBR, but this may be because the silence/music same file size (I did this test and again possibly ABR) and no "VBR" switch in iTunes. I think the white papers and a bit reservoir sited pretty much say AAC's are not CBR. Call it what you will, but as there are many AAC encoders (check over at Hydrogen, I've only used iTunes/QT) out there perhaps someone will do with AAC what LAME has with MP3. MP3 have gotten better, perhaps AAC will to. All tuning. For the record I'm actually preferring AAC's lately, though didn't at first.
 
Oct 20, 2003 at 6:08 PM Post #13 of 25
Well on first listen, AAC sounds better. They seem to have slightly better dynamic range but their soundstage is off. The vocals just don't sound right to me which can be irritating.

I will not have a definate answer until I do the comparison using my ety ER-4s and the amp. Maybe I should have waited to post but since iTunes just came out, I thought it would be helpful to post sooner.
 
Oct 20, 2003 at 7:46 PM Post #14 of 25
iTunes just came out... for the PC.
wink.gif


Actually if you're curious about my initial feelings (with a different equipment than I have now) when iTunes started supporting AAC on the Mac, they are here. Again I no longer completely agree with what I posted. I think there are differences between the formats and some may prefer one over the other. Ogg is often discussed as having a unique signature/problems also.
 
Oct 20, 2003 at 8:52 PM Post #15 of 25
Yes, iTunes for the PC, which is all I have to use. I checked your link and see that there is a quicktime AACelerator. I will have to see if the pc version has this feature too. Maybe that will make the difference. Even if I find AAC format that beats mp3, I have already encoded 60 GB of music in mp3. I really don't think the difference is big enough for me to re-encoded my whole library but I probably would select a few choice cds to do this to and start using it for my future encodings.

Original I had 30 GB of music encoded at 192k mp3 and decided it was worth is to re-encode the whole library at the higher rate. I have no intention of doing this again.
 

Users who are viewing this thread

Back
Top