Previous Entry Share Next Entry
xiphmont

Why 24-bit/192kHz music downloads make no sense

(by Monty and the Xiph.Org community)

Articles last month revealed that musician Neil Young and Apple's Steve Jobs discussed offering digital music downloads of 'uncompromised studio quality'. Much of the press and user commentary was particularly enthusiastic about the prospect of uncompressed 24 bit 192kHz downloads. 24/192 featured prominently in my own conversations with Mr. Young's group several months ago.

Unfortunately, there is no point to distributing music in 24-bit/192kHz format. Its playback fidelity is slightly inferior to 16/44.1 or 16/48, and it takes up 6 times the space.

If you just said 'Whaa?', you may want to read the whole article.

It's fairly long... but hearing, perception and fidelity are complicated topics. Shysters and charlatans exploit that nuance (and misunderstanding) to bilk unsuspecting consumers of their money, all the while convincing them they're paying for 'quality'.

Anyway, happy reading and comments welcome!


Re: If digital is better . . .

dr_memory

2012-03-07 01:08 am (UTC)

"why did I (and, we now learn, Steve Jobs) always find my CDs tiring" This is the real question.

No, the real question is: would you (or Steve Jobs) find the same music selection more "tiring" in a double-blind test where neither you nor the tester know if the ultimate music source is 44/16 or 192/24?

I encourage you to go find the answer and report back, but the paper that Monty referenced under "Listening Tests" strongly suggests that it will be "not more than 50% of the time". (In other words: not at all.)

(Edit: I strongly encourage you to go find the answer for yourself. I do not suggest or condone digging up Mr. Jobs' corpse, and doubt he would make a good test subject in any case.)

Edited at 2012-03-07 01:10 am (UTC)
(Frozen) (Parent) (Thread)

Re: If digital is better . . .

(Anonymous)

2012-03-07 01:54 am (UTC)

A/B testing is meaningless. Extended listening makes the difference, and here's the difference: CDs are irritating to listen to. I can't pin a label on exactly why, and labels describing audio are mostly meaningless anyway - read a high end review to see what I mean. I have a collection of several hundred CDs and I just don't listen to them often. Yes I have done AB tests but it is hard to get the same content on CD/hi-res/vinyl, but when I have the three sources we can tell the difference. (Vinyl is usually pretty obvious, the noise gives it away) But more importantly, when I listen to something that sounds good I'm excited to put on the next thing I want to listen to. When I've heard a CD or two, I just let the CD end and sit in silence. I sit and listen to music for hours a week. This isn't speculation.

I am an engineer and I thoroughly believe in measurements and numbers, but the numbers we think we should be measuring frequently don't correlate to what we hear. I'm also old enough to remember as a teenager the comments about transistor amps being essentially perfect. ("you can't hear distortion under .5%..") I'm hearing same thing today about the Redbook standard and it is simply wrong.

My suggestion to you, if you haven't done this already, is listen to hi-res audio on a high end system for extended periods, and compare it to CD. This means quality electronics and speakers and a DAC that is properly de-jittered and bit-perfect input with properly recorded source. Listen to vinyl on a turntable with a cartridge that costs more than fifty bucks. Then get back to us and let us know what you think.
(Frozen) (Parent) (Thread)

Re: If digital is better . . .

dr_memory

2012-03-07 04:48 am (UTC)

"My subjective experience totally trumps your double-blind tests! If you deliberately avoid blind testing, you'll have the same experience I do!"

Ooookay then.

And there's nothing about A/B or A/B/X testing that requires short-duration tests. You just need a test subject who's willing to put in the time, which it seems like you are. Obviously testing digital vs vinyl is a nonstarter -- as you say, the noise is gonna give it away every time -- but redbook vs 24/196 is just a matter of putting in the hours. If you think you've got the ears to do it, you could make some grad student's day, month, year and career.
(Frozen) (Parent) (Thread)

Re: If digital is better . . .

(Anonymous)

2012-03-07 06:53 am (UTC)

"If you think you've got the ears to do it, you could make some grad student's day, month, year and career."

My ears aren't what they used to be, but it would be a lot of fun to confirm my anti-CD bias. or not. And I've got a couple of friends that would be happy to participate as well. Hell, I'll even pop for a couple bottles of wine... My comment about A/B really applies to short A/B tests which is pretty much all I've seen people do. Blind testing is fine, but it takes some time for a valid judgement and for the general sense of Yuk to set in that accompanies 44/16. (BTW, I agree that 192 is a waste of space, there is some threshold that's Good Enough. Not sure what that is.)

Seriously, I'd be willing to run this test myself if I could get the source to do it. I'm perfectly willing to be proven wrong if that's the case.

What's interesting about this debate is that there are a good number of people that have the same complaint. Seems strange that we'd come to the same conclusion without any basis.

(Frozen) (Parent) (Thread)

Re: If digital is better . . .

xiphmont

2012-03-08 06:01 pm (UTC)

You're not the first person to assert that there should be extended tests. I'm pretty sure I know what the results would be from a BAS style 'insert an A/D/A into the loop' test would be, but there's no harm in doing it...
...just the time....

In any case, some folks are talking about setting up a week-long listening demonstration and test at SXSW next year. No idea how serious they are, but I'm seriously considering it.
(Frozen) (Parent) (Thread)

Re: If digital is better . . .

(Anonymous)

2012-04-28 03:58 am (UTC)

If people claim to notice a difference during normal listening, but fail to notice the difference during an A/B test - there are two possible conclusions. That the A/B test is correct, and that the people imagined they heard a difference that did not really exist, or that the A/B test is not valid. Unfortunately because the A/B test is conducted according to scientific principles, and hence carries the cache of being "science", it is assumed to be correct. If you like the idea of the A/B test, then I guess its easy to say that the test is right and the golden-eared people wrong, or like me you can examine the tests in an open and objective fashion and realise that they are not in fact perfect and are limited significantly by the heisenberg principle - that is that the process of testing itself can alter the thing being tested. They go in the right direction but dont go far enough to be truly reliable.
Gotta love these Captcha's - i keep thinking i should get a prize every time i get them right.
(Frozen) (Parent) (Thread)

Re: If digital is better . . .

xiphmont

2012-08-05 11:52 am (UTC)

The Heisenberg principle explains the duality of waves and particles. It's a subatomic phenomena, and it's very real (not simply a thought experiment).

The principle has been lifted by metaphysicists as an analogy to describe uncertainty in the macroscopic world. It's a powerful idea, but it's not the same thing when you're talking about people (and not quarks).

Just because it's conceivable on an atomic scale that all the Brownian motion in a pot of water will cancel out and the water will spontaneously freeze, it doesn't mean it's actually possible. Macroscopic systems are governed by by the same rules, but they manifest in completely different ways.

Heisenberg does not explain why silver-plated pants cannot keep my ass from looking fat, no matter how expensive they might have been.

(The pants, not my ass.)
(Frozen) (Parent) (Thread)

Re: If digital is better . . .

xiphmont

2012-08-05 11:56 am (UTC)

(I should say though--- your point that testing cannot be perfect is true! However, testing can be, and usually is, very good. Don't forget, we're not asking the crazy gear junkies to win most, or even some of the tests. We're waiting for them to win _any_ of the tests. If you crash your car into a tree every time you take a driving test, you don't get to blame it on the instructor stressing you out... or rather, I suppose you can, but you still don't get your license)
(Frozen) (Parent) (Thread)

Re: If digital is better . . .

(Anonymous)

2012-04-28 03:40 am (UTC)

I agree with you about A/B testing being meaningless, for so many different reasons. For one thing I believe sound perception is influenced to some extent by the frame of mind or mood the subject is in at the time, and there must be a great difference in frame of mind between normal listening and when the subject is taking part in a test. I would drive very differently when doing a driving test than I would when i was driving alone. To conduct any sort of meaningful A/B test you would have to do it in such a way that the test subject was unaware they were being tested - good luck with that one.
(Frozen) (Parent) (Thread)

Re: If digital is better . . .

xiphmont

2012-08-05 11:43 am (UTC)

This is, frankly, a complete cop-out.

The average audiophile rag typically describes differences as 'night-and-day' or 'a whole new world' or similar hyperbolic terms. These differences reliably evaporate in the presence of witnesses and rigor.

...nor do I think you'd get very far explaining to your driving-test examiner that you only get into multi-car accidents when being tested, but when you're on your own, everything is always completely fine.
(Frozen) (Parent) (Thread)

Re: If digital is better . . .

(Anonymous)

2012-03-07 10:38 pm (UTC)

Is there a practical test that people can do at home?
Maybe this?:
Rip CD to FLAC.
Compress FLAC file to MP3 with bit rate = 48kHz.
Using same encoder, compress FLAC file to MP3 with bit rate = 192kHz (or anything higher).
Have someone else choose the files to play. Same stereo, same song(s).
Guess which bit rate you're hearing.

Would this be a good test?
Why not?

THANKS!
(Frozen) (Parent) (Thread)

Re: If digital is better . . .

xiphmont

2012-03-08 06:03 pm (UTC)

No. The CD Starts out at 16/44.1. Upsampling adds nothing. All you're really comparing is that mp3 encoder to itself (and perhaps the quality of your software resampler).
(Frozen) (Parent) (Thread)

Re: If digital is better . . .

(Anonymous)

2012-08-05 11:06 am (UTC)

Why are you so against advancement in audio technology? Don't look for ways to justify your reluctance for better audio. If all scientist felt the same way there would be no moving forward in the 21st century.

The reason for music being better than cd on 24/96 or 24/192 is not so much the increase in frequency but the increase in bit rate depth. There is a lot out there that we still don't know and trying to justify it all with limited scientific measurements (just because at present we do not measure or can not measure these effects properly) is simply narrow minded. Don't be a Luddite.

If you want to want to hear top quality audio, the best and most satisfying way is still via a good quality analogue turntable, amplification and speakers playing original vinyl recordings.

No one made the link between the down turn in music sales and the start of cds in the 80's but this is the reason. Surely we didn't all just accept it without question when Sony and Philips told us it was bit perfect digital audio? CDs strip away the soul and emotion of the music and yes long listening sessions are fatiguing (it is a compressed format). Then in the 90s we take this soulless form of audio and compress it further to take the last remaining sembelance of music completely away (mp3) and offer it to the public as music! Now we have the chance and technology to rectify this 30 year recession in playing music and there are still dissenters. There are many people who in the 21st century have never heard hi res audio (proper vinyl or 24 bit recordings) and its a shame and many of these people think mp3s sound good only because they don't know better.

Lets move things forward not seek to stifle them.
(Frozen) (Parent) (Thread)

Re: If digital is better . . .

xiphmont

2012-08-05 11:24 am (UTC)

Far from declining, music sales by both dollar amount and number of albums moved peaked (by a considerable margin) during the CD era.

The rest of your statements are similarly ignorant of the basic facts. Sorry, there's simply no way to sugarcoat it.

Do go on enjoying vinyl; no one's suggesting you shouldn't. It was the best we had for over fifty years. Claiming it's higher fidelity than digital though-- I'm sorry, that's very simply wrong by every measure of wrong.

I'm rather fond of classic Studebakers myself, but I'd never claim one could take an M3 under any circumstances.

Edited at 2012-08-05 11:29 am (UTC)
(Frozen) (Parent) (Thread)

Re: If digital is better . . .

dr_memory

2012-08-05 04:17 pm (UTC)

Consider reading the original article that started this thread.

Also: "this thing is better in a way that I cannot quantify or measure but you're a luddite for not believing me" is the most hilarious argument ever.
(Frozen) (Parent) (Thread)

You are viewing xiphmont