This all hinges around a supposition of how you’ll interact with digital recordings that you have paid for in the future.
I have about 800 CDs in boxes under my bed now and a bit more living space in home, which I quickly filled with a new turntable and all my old vinyl.
I am worried though that when my descendants all have genetically engineered super-ears, that great granddaddy’s record ripped record collection is going to sound like a load of old wax cylinders!
Couldn’t complete the survey, sorry (two minutes of that song is enough for a lifetime 
). First sample sounded quite good actually. But to equalize (unconditionally) the bitrate, judging by only one song (or even one genre), is a bit of a stretch, don’t you think?
PS What’s wrong with ‘Rikki don’t lose than number’ http://www.soundonsound.com/sos/sep03/articles/testcd.htm ?
I also don’t think this is a very good way to test the quality of the encoding. Since we don’t have a reference (the uncompressed CD audio) what you’re really testing is if any particular encoded version sounds the most appealing to the majority of people. That is not the same thing as an accurate representation of the original. Say for instance you would produce a truly horrible and annoying set of sounds. The encoding of that may remove some of the frequencies that are causing it to be horrible, and people would then rate that version higher. But it’s not what you originally intended.
A better way to test this is, in my opinion, to give us the uncompressed version to compare against. Then the uncompressed version would be among the compressed ones we compare with. Then you would see if people really could tell a difference or not.
@Jeff: I must say I’m very surprised that you, of all people, would post something like this? You where the one to blog about preserving everything, and yet you want to throw information away?
This is actually the first time I’ve read a blog post by you that I really disagree with, for reasons stated multiple times by other commenters (Brent Maxwell and Ben Leggett just to name two), but oh well, nobody’s perfect. 
Thank you for a very great blog though, and it’s actually the only one I recommend others to read, but do so a great deal, because of the very high quality of posts, and all the information you manage to cram into them!
My kids listen to vinyl and say it has nuances that CDs and MP3s don’t capture. Like clicks and pops I can only imagine.
This is actually the first time I’ve read a blog post by you that I really disagree with, for reasons stated multiple times by other commenters (Brent Maxwell and Ben Leggett just to name two), but oh well, nobody’s perfect.
I wholeheartedly agree with Johny Woller Skovdal’s comment.
Henrik was the first person to get this right.
Ripping lossless isn’t really about audio quality, but rather generational quality.
If I convert my CD collection to 192kbps MP3 format, I’ll be very happy to listen to it… for as long as MP3 remains a viable format. Given that (a)it’s patent-encumbered, and (b)technology very rarely regresses, I can imaging wanting all my music in a different format at some point in the future.
Transcoding a 192kbps MP3 to something else causes another round of loss – one you can definitely notice. Transcoding the new file to a third formation would be yet another round of loss; and the quality degrades rapidly with each generation.
As you yourself point out (https://www.google.com/search?q=site:codinghorror.com+hard+drives+cheap), hardware is cheap; including storage. Why would you not make yourself a lossless copy of your collection somewhere, and then carry around your lossy files for listening to?
MP3 = LOSS at any bitrate. 320 kbps is great for earbuds - that’s it. I will not waste time comparing cds to mp3. There is none. Buy vinyl and a decent table if you want to hear the recorded track in it’s purest form. Until you jarheads start MAKING digital music (and some crap I hear these days sounds like 2 notes over and over… 010101010) you will always need to convert to analog to hear it.
Go ahead and settle for micro speakers and video on phones. Bigger is better.
1: Limburger - Sounded Nice (Could be uncompressed)
2: Cheddar Sounded worse than Limburger
3: Gouda sounded OK
4: Brie Sounded Nice (Could be uncompressed)
5: Feta Sounded the worst of the lot.
2,3, and 4, were pretty similar to me. I don’t know if I was imagining things or not. I tried not to put too much thought into it, but just give my impressions. I am an audio engineer, and not an audio snob. I want the best audio of course but if I can’t hear a difference it’s not biggie to me. I want of course to archive music with an uncompressed format of course. I encode all my MP3s of my music as LAME. I have to put most stuff up on Reverb nation at 160kbps due to size constraints, and I thought I heard a hit in quality over my 320kps files. They sounded more dull. Now that I think of it, I am exporting recent MP3s out of the Cubase project, rather than using WaveLab as I used to. Hmmmm. Maybe Cubase isn’t doing LAME?
Chad Johnson
That was a really interesting post, and a really fun test to boot. Good timing too, now that my audio system is setup
I really hope you aren’t pulling our legs though and they all end up being exactly the same… By the way, are you going to normalize for stockholm syndrome?
Woof, woof, woof. Woof woof woof, woof woof!
@Tristan Harward: That sort of misses the forest for the trees. You’re saying the sound quality is better, even if the human ears cannot distinguish between the two. But what’s the point of having better unless you are going to be listening with something other than your ears?
Woof, woof:
Best = 5, Worst = 1;
WOOF WOOF WOOF WOOF WOOF!!!
I totally agree that decent-quality MP3s are “good enough” for listening, even on decent equipment. I tend to go for 256kps on average myself, just because “hey storage is cheap”. However, to me that’s still not argument for ripping straight to MP3. The point of ripping your collection to FLAC is that you can toss the CDs and never worry again, forever. You have the full fidelity original recording digitized faithfully, and you can then encode that to MP3 or any future standard at-will.
Re-compression with multiple codecs does degrade quality. If you rip to 192K MP3s and toss the CDs, in ten years (or more!) you might decide you want your music collection to be in 128K MZ6 format, because it’s even smaller and better sounding, and one of your newest listening devices has dropped support for that old MP3 format. With a FLAC library you can do that. With only your 192K MP3 library you’re hosed, you’d keep losing quality every time you transcode to a new format.
My recommendation, if you want to keep digital audio around at all: Rip to FLAC for archival purposes. You can put those on slow drives or backup media or whatever suits your fancy. These are your permanent archives. Then encode the FLAC data down to your MP3 flavor of choice for your live library that you actually share around the house and listen to on devices/computers, or upload to a music locker service, etc.
I guess you made your point about having really-good audio equipment.
I think that the sample chosen for the experiment is biased towards the point wanted to be made. I pass.
I believe it’s quite hard to differentiate it -because we have no base ground on how he actual sounds are, when uncompressed- that’s not how you do AB tests.
If you want to challenge us about sounds difference, it’s better if you give us the very same track, compressed on different bitrate. Not different song, which may or may not bad recording from the very start, this can lead to different results.
The Great MP3 Bitrate Experiment may be worth doing with classical music—the results of reproducing a 100-piece natural-instrument orchestra (especially if live, not postprocessed/edited significantly) may be more striking than a small band with much of the sound electronic (or electronically post-processed).
Another poster was right—cymbals are a way of distinguishing quality. Is it live or is it Memorex? You can tell with cymbals.
I am actually surprised by the advances in modern MP3 codecs and disappointed by aging of my hearing. Many years ago I performed non-blind test and I could easily hear the difference between CBR 160 and CBR 192; and I believed I could distinguish 192 and 256.
But in this test I don’t hear any real meaningful difference (old Audigy + AKG K390). Will try next week with high quality sound card and good headphones.
I went and downloaded Carmen samples from here http://gizmodo.com/5251247/the-great-mp3-bitrate-test-my-ears-versus-yours
I renamed files randomly. And still I could guess 64, 96, 128 and 320(!) bitrate exactly in that order on K390 headphones. Maybe it’s a bad codec in this case but I am staying on 320 bitrate because I can easily hear the difference.
Well then. They all sounded roughly the same (Baad) except for one which was terrible.
I’d be interested to see a comparison like this done with some Black Metal, one live track comes to mind: Satyricon - Mother North (Live from Gjallahorn).
It features a brass section, crowd noises (chanting), slow drumming, fast drumming, even faster guitars.
Here’s a crap version on YouTube: http://www.youtube.com/watch?v=1gye4XCFAmo
