Hi all.

I'm planning to get a 160GB iPod Classic around the end of the year, but I want to start ripping CDs with iTunes now, as I have about 1,600 discs that I want to transfer. I've estimated that the highest quality I can use and still get all, (or at least most) of these files to fit would be AAC 256 kbps, with VBR encoding turned on.

In an ideal world, I'd first rip everything to Apple Lossless and then covert those files to AAC for the iPod. However, storing this much music in both formats would use up more disk space than I can afford right now, since I also want to keep a second copy of everything on a backup drive. Still, I may rethink my plan if I'm going to take a significant hit in sound quality. I should note that I'm only going to use the ripped files for portable listening, and will still be using my CDs when listening on my home system.

So, my questions about all this are:

1. What are the audible artifacts of lossy compression when using higher bit rates like 256?

2. I’ve heard that AAC is superior to MP3 at lower bitrates, but what about higher ones like 256 or 320?

3. What differences would I be able to hear between files encoded at AAC 256 VBR, vs. Apple Lossless if using a decent set of earphones? (I’m currently planning to get a pair of Shure SE210s with the iPod).

Thanks.

1. We can't really answer this. Different people hear different things with different audio formats. You have to answer this question yourself. Take a blind ABX test using software like foobar2000. Generally speaking though, when using high bitrates like that, it is damned hard to notice any audio flaws as modern day lossy encoder can perform very well.

2. Again, you have to determine this. My experience has shown that (for me) the Nero/iTunes AAC and Lame mp3 formats all sound the same once you get up to bitrates at around 190kbps VBR and above. Then again, this is just with my ears.

3. Don't really know, I don't have your ears.

I would recommend using the Nero AAC encoder instead of Apple's, if it all possible. The iTunes encoder has some bugs, even at high bit rates.

ABX your way through, you will probably notice that you dont need such high bitrate as 256. Waste of precious bits!!

For me itunes at 128 is transparent (well atleast in 99% of the cases)

Happy ABXing ^^

....on those all-important thunderstorm recordings

Was that the issue with stereo imaging? I thought those were fixed.

The previous answers are technically correct - we don't have your ears - but we can reflect the combined experiences of many people and many listening tests as follows.

1. Unless you have a really bad encoder, artifacts are few and far between at 256 kbps. There are killer samples, but most of the time most people won't hear any difference from the lossless original (i.e. it is transparent) under proper blind test (ABX or ABC/HR) conditions. For a highly-rated encoder like iTunes or Nero for AAC-LC or LAME for MP3, you'll be fine for the vast, vast majority of your music.

2. From around 128kbps, the LAME v3.97 (-V5 --vbr-new) MP3 is almost as good as iTunes AAC, and both are widely considered to be mostly transparent or most differences that can be perceived are not annoying. For most people LAME -V3 --vbr-new (~160 bps) is pretty robustly transparent and LAME -V2 --vbr-new (190-200kbps) is very robustly transparent. For compatibility (e.g. if you get a car stereo that supports MP3 but not AAC) you might consider using MP3 as it remains by far the most widely supported lossy compression format. If you're sensitive to pre-echo, you might find that AAC's improved time resolution is an advantage over MP3 at any bitrate. You're very unlikely to hear much difference between any of the good encoders once you are above about 160-170 kbps in VBR mode, and there's little to be gained as you go even higher (e.g. 256 or 320 kbps).

3. You're very unlikely to be able to tell the difference between any good lossy encoder (including iTunes) at 256 kbps and the lossless original for the vast majority of music regardless of your earphones. For that matter, 160 kbps or 192 kbps is already at that point. With loudspeakers, it's even harder to tell the difference than with earphones/headphones.

Excellent points Dynamic. From a practical point of view there is little gained going over 160 k with these encoders. Currently I use Lame 3.98b -V4 and see very little benefit in going higher than -V3.

Thanks Dynamic, and everyone else who replied. It sounds like I'll be good at 256 VBR. I may even try some tests at 192 just to see if I can hear a difference. I do have Nero 6.6.1.15, but with all the CDs I need to rip, I think I'll stick with iTunes for the convenience. Fortunately, I don't have too many glockenspiel or thunderstorm recordings.

Since I didn't see it above I figured I'd chime in and suggest taking your listening environment into account when choosing a bitrate. While you may be able to tell the difference between 128 AAC and 256 AAC while in a quiet room at home (and even this should be difficult), you probably aren't going to be able to do so quite as easily somewhere on the move where there is significant background noise.

I've had good ripping my CD's to WavPack images for home listening and archiving purposes, then batch converting them to 128 AAC VBR using foobar, iTunesEncode46 and iTunes for on the go.

I also remember a series of threads in the past that suggested smaller files sizes would lead to a longer battery life... although I don't remember if this was proved or disproved, you may want to look into that before ripping everything as lossless for on the go.

Until you acquire some new music but have no space left on your 160G iPod because you used an unnecessarily high bit rate.

I just preformed a dual rip of a dozen CDs into my Vista Laptop and from it into my 3rd Gen Nano to answer your first and third question:

The artifacts are a subtle but noticeable in even in portable environment. With Lossless the high frequency transient notes have a sharper/faster attack while at the same time a slower more complex decay. This allows instruments to come forward and have a separate sound. With midrange the Lossless adds warmth (mid-bass?) to female vocals and guitar chords are heard as individual notes rather than a blended sound. Low frequencies are less muddy and more notes are individually separated.

Whether this is worth the 3X increase in file size depends on how disproportionate the emotional affect these subtle changes have on you. I was surprised how much more I enjoyed the Lossless file. Also the AAC@ 256kbps was more fatiguing to listen and I found I myself lowering its volume while raising the volume on the Lossless file.

Note: Listening was done in a car with half of the time using Ultimate Ears super.fi 5 Pro earphones and the other half the factory stereo on PT Cruzer (a rental) with a cable connection.

Please see TOS #8. The idea that you could discern lossless from 256kbps AAC on a Nano via the analog out into a PT Cruiser's aux in is absolutely absurd.

I demand ABX logs.

on topic,
256kbps should be plenty. I agree with the others, see if you can ABX lower bit rates like 160 or 112. My personal preference is to always go as low as you can go Why waste space?

I second this one. It really pays out to work as compactly as possible, hence I go for codecs which offer decent performance at low bitrates. While I and many others (referring to personal ABXing and recent HA multiformat tests' 96 kbit/s high anchors here) are unable to discern Nero AAC -q 0.34 files from the originals, these encodings are small enough to fit on any flash-based portable device. Friends of mine have become quite used to asking me to encode their CDs to this format, so they can store them on their AAC-capable mobile phones. Mainly because these ones suck less space compared to their own Media-Player-MP3s. LAME MP3 isn't exactly a codec optimized for low bitrates, as Gabriel Bouvigne also stated in older threads about this topic.

Edit: I want to add that I intentionally abstain from training my hearing to find subtle encoding artifacts in my music, therefore bitrates generated by aoTuV b5 -q2, Lame -V5 and Nero -q 0.34 are perfectly fine for me. I rather prefer listening to and enjoying the music, instead of searching for annoying flaws in it.

In addition to the above comments, the Audio Engineering Society has a test disc specifically designed to show MP3 compression artifacts and to educate people as to how to hear them.

You can get the test disc for about $20 at the link below:

http://www.aes.org/publications/AudioCoding.cfm

Too bad they don't provide this CD as a download!

--Vidiot

Did you ABX that?

I made similar mistakes once (long time ago.) I thought I could easily differentiate between 192 and 256 kbps mp3. Once ABX'd I realized that I couldn't.

I think Nero AAC encoder is much better than any mp3 encoder at all bit rates.
As to how to choose the encoding bit rate - a lossless encoder generate something around 700 kbps,
NeroAACenc - around 400 (the max) kbps with (quality factor) -q 1.0 and around 320 kbps with -q .75.

So I take it that you have compared Nero AAC against Lame mp3, iTunes mp3, FhG mp3, Helix mp3, and every other mp3 encoder out there using every bitrate and VBR encoding schemes with blind ABX methods. That must have taken you some time to come up with the conclusion that Nero AAC is better than any mp3 encoder at all bitrates. I would sure love to see those ABX results.

Using Nero AAC at -q0.95 (~400kbps VBR) is rather high given that a 4 and 24 second file comes out to about 12.3MB in file size. I think there is no need for a file of such high bitrates given that the quality increase between 320kbps and 400kbps isn't worth the extra gain in file size. That and if you are going to use 400kbps you might as well go all out with lossless.

stereo_dom...

Total Cumulative Posts: 1
Last Active: 24th October 2007 - 19:43

(at the time of this post)

...I hope you guys aren't actually expecting a reply.