source comparisons - mp3 and CD

[andrzejpw]

Quote:

Originally posted by JaZZ kerely... ...good analogy with the screen resolution! Actually I can set (and I do) the zoom factor to 100%, so I see every single pixel and definitely the data reduction artifacts, if they're obvious enough. But let's be generous – your analogy is very appropriate, I think. Well said. You just have to add that the same applies to solid state equipment. So I'm not so sure if a tube amp is definitely worse for critical, comparing listening. My EMP is as adequate as my SS amps.

Jazz: Its not the zoom that counts! Its the resolution. Zoom in on that jpeg. Yup, zoom in a bit. See that? Now, print the tiff and jpeg out. See THAT?

 

[Blighty]

Quote:

Originally posted by JMS The fairest way to evaluate the MP3 encoding process that consumers can readily perform, in my opinion, would be to extract a CD into WAV, encode it using MP3, decode it again into WAV, and then burn BOTH the original WAV and the extracted/reextracted WAV onto CD. This minimizes distortions introduced by the CD reader/burner, differences in the playback device, etc. I have done exactly this using various tracks (mainly classical, but includes vocal and choral), encoded and decoded using MightyDAC's LAME MP3 codec at 128kbps, set for "best quality". I used the disc to conduct some brief blind listening tests on a proper stereo with 3 of my friends, who are not audio enthusiasts. I had tell them which was which without revealing the source. In summary, everyone concluded it was damn hard to tell the difference, as they all mainly guessed for their answers. I have listened to this disc extensively myself and agree with their findings, although I THINK I can tell a difference on a track or two now, but I would have to confirm it with more blind testing. Now, I have played this disc on various speaker/headphone systems, including an Arcam Alpha 7SE -> META42/AD8620/2xEL2002 -> Sennheiser HD580 configuration. If we had so much trouble telling 128kbp's MP3's from the original, I wonder how everyone on the internet can tell a difference with 192kbps by playing through their computers in a noisy computer environment? If you still don't believe me, there is an article published by an audio mag (I think Stereo Review) a few years ago that basically concluded similar results. I'll post when I dig up the URL in a bit.

I have done this test before under the proper conditions you have cited. I could tell the 128 version from the 192, 320 and original WAV file every time with careful listening. 128 is not by any means bad but I would prefer at least 192 which seems like the threshold where it becomes statistically meaningless in attempts to discern it from the original WAV file. So 320kbps MP3s are definitely more than sufficient quality for just about every one. It may still be technically lossly but I seriously doubt anyone can tell 320 apart from a WAV in a properly controlled DBT.
The audio industry is full of frauds and con artists perpetuating myths that often mislead naive newcomers entering the hobby while continuing to prey on a great deal of audiophile veterans who practice the art of subjective evaluation exclusively.

 

[NewportGeek]

look, i'm like 3 months away from my computer science degree, so bieleve me when I say that i know a fair amount about compression algorithims.

MP3s compress sound using rules people have come up with on how our brains intrepret sound. (like if there are two sounds being played simultaniously we hear the louder of the two.) Why not just use a CD player and get your sound free of some programmers subjective analysis of how your ear hears?

You do realize that the whole MP3 argument is that MP3s are equal to cds. No one is saying that MP3's can soud better than CDs.

That means the best you MP3 folk can hope for is equality. Why not get a decent CD player and know that MP3's will not sound BETTER than your cd's. In other words MP3 <= CD.

I know all of you MP3 folk are saying EQUAL EQUAL. Well fine, thats an agument for another day. All i know is my CD's sound amazing, and my SACD's will leave you in the dust.

-JT

 

[JMS]

As, promised, the MP3 vs CD articles. There are actually 2. Basically, they did a first one with audio enthusiasts in an "audio club" and these people can't tell a difference, for the most part. Then they redid the test with a few very select people who are audio compression specialists, they could more or less tell a difference.

First article:
http://www.soundandvisionmag.com/Sou...2020_1,00.html

Second araticle:
http://www.soundandvisionmag.com/Sou..._cm_00,00.html

 

[aeberbach]

I finished my CS degree about ten years ago and back then I thought I knew everything too. What you learn at college is how to learn and very little else.

I have a great CD player, Arcam CD23. I have all my CDs encoded to MP3 using LAME 128-320k VBR and there just isn't any practical difference. I still grab the CD when I want to really listen to a CD just because I like my CD player. If space were more limited I would be just as happy to play my music on a computer into a good DAC and amp.

The statistics and the tests don't support the idea that CD is necessarily better. They say that most of the time CD is indistinguishable from MP3. In the soundandvision tests, for everything but the solo clarinet most people didn't have a clue and even then many didn't pick it correctly.

It would be really interesting to try and duplicate this test. Does anyone have enough time to prepare all the samples as the article describes, distribute them and gather results?

 

[KingNothing]

although off topic, my solution: APE it.

 

[NewportGeek]

aeberbach : where did you goto college and not manage to learn anything?

seriously, i don't think i know everything. But do know quite a bit about mp3s. I stand behind everything I've said.

CS 10 years ago is not what CS is today.

just for the record.

-JT

 

[taoster]

the other problem i have with computer based systems is the machines. they are noisy with the fan and when absolute detail or silence is required, it dont work.

 

[JaZZ]

Quote:

Originally posted by andrzejpw Jazz: Its not the zoom that counts! Its the resolution. Zoom in on that jpeg. Yup, zoom in a bit. See that? Now, print the tiff and jpeg out. See THAT?

No, I don't. 100% = 1:1, so every data pixel is corresponding to a pixel on the screen – result: no interpolation, full resolution, although there's possibly just part of the picture to be seen on the screen.

 

[taoster]

Quote:

Originally posted by JaZZ No, I don't. 100% = 1:1, so every data pixel is corresponding to a pixel on the screen – result: no interpolation, full resolution, although there's possibly just part of the picture to be seen on the screen.

that's a good example really.. a 320*200 screen maybe its transforms to exactly the same but on a 21" running at 2000pixel+ the difference is obvious.

 

[aeberbach]

Nothing stays the same but the great thing about experience is that you stay current and keep learning. You'll know that when you have some.

How about you, NewportGeek, run the test? Prove your claim.

 

[austonia]

This coming from a guy with pretty ordinary set of ears.

I used to buy CD's, a lot. I have about 250. They do sound great, I have no complaints on audio quality. My complaint was on price and durability. CD's @ $15 and sometimes you get buyer's remorse since you coudn't check them out before purchase. Also they occasionally get scratches. A minidisc-like housing would have been so much better. My music has to be on the go so they can't always be sheltered.

Then MP3s & Napster. Between my own rips, MP3 swaps, downloading, etc I have hundreds of hours of music I havn't even listened to yet.

MP3's offer NEAR CD quality, portability, replacability, and easy sharing (did I mention...Free). I love it. You can carry 1000s of songs in your pocket instead of lugging around my CD binders. Some MP3s do sound poor and of course thats an issue mostly with the encoding process, but you can always find that same track with better coding if you poke around. I am sure true audiophiles prefer CD (SACD, DVD-A) but if you can step back just a little then you gain so much. The tests made between CD & MP3 show little difference except to the most trained ears. Note I only count 192kbs + as "high quality" the rest is more like radio (128kbs or less). You can tell from the amount of hardware out there its really caught on with consumers.

I still buy a few CD's in the last 3 years to supporting a artists who deserve my love. I would consider buying more at CD < $8

I run a Nomad Zen 20gb -> FM remote -> Koss PP's and it sounds very decent for a portable rig. I have not even cosidered going back to PCDP until i come to this forum.

 

[Vertigo-1]

If you know what to listen for, it's actually not very difficult to distinguish a 192 MP3 from a WAV file, even off a crappy soundcard. Using my Soundblaster Live, and Rotel integrated driving a pair of Grado SR-225s, the WAV file contains an airier sound...a major key component that high end CDPs have over low end CDPs, and a key component that good CD playback has over MP3s. The MP3 of the same song sounds flat, and lacks any sort of 3 dimensionality, or depth. Lifeless is a good way to put it. The average person probably wouldn't know how to listen for this difference...the common reason why some people think all CDPs sound the same. They're out looking for tonal differences, when differences actually lie elsewhere such as in PRAT, soundstaging, resolution, or depth.

 

[SumB]

I didn't have difficulty discerning between 320kbps LAME encoded mp3's and the EAC ripped WAV's through the crummy DAC of a SB Live.

As Vertigo stated the changes are to the subtleties the average person isn't listening for, such as the frequency extremes (softened deep bass and compressed, grainy highs), truncated dynamic range, ambient detail (air, depth, shimmer, decays, natural reverb, subtle low level sounds), transients, instrument separation, and so on.

All it takes is a keen ear and some well produced music one is familiar with. This test debunked itself by using unknown participants sampling unkown music in a genre with the likelyhood of being highly compressed and devoid of dynamic range.

 

[JMS]

Were your evaluations done under proper conditions, as mentioned above? Did you do a blind test?