A. Some DACs exhibit lower distortion, etc at one or more higher rates. Therefore playback at the higher rates will get you a cleaner output. Miska-Jussi of HQP has documented this.
B. Different/less steep filtering available with hi res may sound different/better to some users or on some systems. It also allows aliasing artifacts to be pushed well above the limits of human hearing and be filtered out, which doesn’t necessarily happen in Redbook.
C. Hi-res versions of albums often have a different mastering than Redbook, mp3, or even CD download versions of albums. So that can be a reason to buy them. One of many examples: the 50th anniversary remaster of the White album: the BluRay stereo version in 24/96 has a different sound than any of the other versions - even the 24/96 download/streaming version. There are lots of hi-res versions of albums where the mastering is deliberately different, done to appeal to an audiophile concept of how a recording should sound - for instance mastered with less volume compression-and not tailored to sound good on earpods-as opposed to CD or mp3 versions.
A. Totally agree. I would argue a good DAC should handle all formats transparently, but software DSP is always more capable. However, up-conversion can be done at playback time and can be configured according to each DAC. That is the reason Roon supports that (and one reason why I like it). It makes no sense to keep redundant data in the library at all times.
B. Why is aliasing an issue? A/D conversion uses oversampling, so analog filters can roll off slowly, while rate down-conversion is performed with digital filters, which again can be pretty steep without objectionable artifacts. Redbook’s out-of-band region between 20kHz and 22kHz is enough for 16 bits, no need to go beyond that.
C. If CD versions are deliberately mastered differently from the hi-res versions in order to make them sound worse and justify charging more for hi-res, isn’t that the definition of scam? If instead of publishing a well-mastered 24/96 version, they just down-converted it to 16/44 and publish that instead, wouldn’t it sound the same? If excessively-processed and compressed versions are by design, why are the versions labeled “hi-res” and “CD” instead of “airpods” and “audiophile” (or “original” and “dynamically compressed for your convenience”)? Because the narrative is that the format alone makes the difference, and that is the scam.
A. Yeah, and not everyone uses Roon, does that, etc. In addition, some people just like to have the file in the format it was produced at, if they can.
B. Maybe aliasing shouldn’t be an issue, that doesn’t mean it isn’t in the real world production of a CD or an mp3. There are all sorts of reasons a file could have aliasing artifacts. And analysis of files shows they sometimes do. Once it is in your file you can’t get it out. It’s just a actual reason to record and playback in hi-res.
C. I didn’t say CD versions were deliberately mastered to sound “worse”. I said they are sometimes mastered differently. That could reasonably take into account that the final use scenario is earbuds/mobile audio. Part of mastering is tailoring the sound for it’s final use scenario. If that’s for mobile audio, nothing wrong with that per se. If you call that a scam, then it’s the CD version that’s the scam, not the hi-res version.
You give way too much credit to record labels and distributors. In most cases the last thing they care about is informing the public. And I don’t deny that there are many cases where the “hi-res” version assumed to be audiophile isn’t.
There have been a few releases labelled as compressed vs audiophile, by the way.
I’m old, so my experience is rooted in the days of the 1970’s. Mastering for different formats had always resulted in different in audio perceptible differences in the sound files, copying the 1st generation masters to multiple 2nd and 3rd and 4th etc, generation masters. These in turn used to as the source for further mastering for consumer formats, the loss of information between the 2nd generation copy and the resulting lp / cassette / cd was informative & alarming in the days of analogue. But I was also surprised to hear audio changes such from the duplication of 2nd to a 5th generation tape. Loss of harmonics in the curtailed formats, throwing away information on the basis that it should not in theory be audible as it’s too low level or again not in theory be audible as it’s masked in volume in the 20Hz-20KHz range is not sound. Science is a means of producing models to understand the world, models come and go as more information becomes accepted fact, overturning previous models, so be science is routed in statistical analysis because there is no way to test and validate in the classical sense, with so much yet to be discovered about our wonderful hobby of audio, using science to close off thought, further experiments and a future better model of audio perception is unscientific! So Danny I’m with your line of thinking.
I don’t think we can apply analog experiences to digital audio. In particular, analog generation loss is mainly caused by the accumulation of noise and distortion during replication and has nothing to do with “throwing away information on the basis that it should not in theory be audible”. (One of the benefits of digital is doing away with generation loss.) And there is a way to test audio quality, independent of any psychoacoustic models; it’s called ABX testing. It may not be perfect, but until somebody comes up with a more accurate, non-biased way of evaluating musical qualitative differences, it is the de-facto standard. It has been for decades now. I am not aware of any proper ABX experiments that confirm any qualitative differences between an original hi-res track and the same track after it went through the Redbook ‘bottleneck’. As far as science is concerned, there is nothing there, regardless of anyone’s “line of thinking”.
http://www.aes.org/e-lib/browse.cfm?elib=18296
“[r]esults showed a small but statistically significant ability of test subjects to discriminate high-resolution content, and this effect increased dramatically when test subjects received extensive training.” In other words, the average person can hear a difference, and people who have invested in carefully training their hearing can clearly hear differences.
There are many such experiments, and as @danny2 says, a common results is that your average guy off the street has difficulty telling the difference but an experienced listener can tell.
Similar to wine. Or whisky. Or tires on a Ferrari. Or violins. Or violinists. Or tea. Or Hahnemühle Rag vs. Epson Legacy Textured paper. Or Leica Summicron 50mm f2 vs. Zeiss Milvus 50 mm f2.
I am aware of that. It is not a test per se, it’s a “meta-analysis”, i.e. a study of tests, so their conclusions are open to interpretation. I read it some time ago. They say “sufficient data could be obtained”, but if I remember correctly, they were not always able to determine exactly the conditions in which those tests were performed. There are other AES publications that attest to the contrary.
Regardless, I think everybody who claims to be able to tell differences between CD and hi-res ought to do the following:
Take a statistically-significant set of hi-res tracks (20-25).
Convert those tracks to 16/44 and then back to original resolution (so that there will be no DAC-specific or any other audio-stack-specific differences). You can use any free DAW for this (e.g. Audacity). Make sure you select highest quality and shaped dither.
Perform an ABX comparison between original tracks and converted ones. You can use Foobar2000 and the ABX plugin at http://www.foobar2000.org/components/view/foo_abx). You don’t need to use ReplayGain since each pair should be equally loud already. You do want to use ASIO though to make sure audio goes directly to the device.
See how many correct answers you get.
Note that for a “statistically significant result”, you need a confidence level of 95%. For example, for 20 tracks, you need to get at least 15 right. Check out https://en.wikipedia.org/wiki/ABX_test.
This will tell you if hi-res makes a difference for you.
Funny you should mention wine. It seems that people almost invariably prefer the more expensive wine when they know the price, but either can’t tell the difference or even prefer the cheaper one when they don’t. That’s why experiments need to be blind.
Would you have the same cynical view on equipment?
Wrt headphones, I’m a long-time Audeze fan, upgraded to their top of the line, then bought the Hifiman Susvara, then the RAAL SR1a. All nose-bleed expensive. I am quite clear on the difference, quite clear on which I prefer (I reach for them most often), and the ranking is not according to price. I do not need to do a blind test.
I have argued in a recent post that if you feel a need to convince others, or yourself, that a difference is audible you should reconsider your priorities. But there are many elements in our hobby that are quite clear to me.
To be clear, my point here is fake Hi Res. You buy a 24/96, pay an extra for that and then you find there is nothing above 22khz, which let you with a standard 16/44 file, or 24/44 in the more optimistic scenario.
I am not discussing the difference between TRUE Hi Res vs standard resolution files.
I am not trying to be cynical, just factual, to the extent possible. And I’m not going to digress and talk about equipment.
Bias is not a weakness, it is inherently human and thus very hard, if not impossible, to eliminate with willpower. It is similar to an optical illusion, where you are convinced you see spirals where there are only circles, or that two patches of color are different when they are identical. Educating critical listening does not educate bias. Everybody needs a blind test. If it’s not blind, it’s not objective and trustworthy. There is nothing cynical about that. Blind-testing headphones is almost impossible; I won’t fault anyone for not doing it. Blind-testing digital tracks on the other hand is quite easy, which is why I personally don’t understand why people resist it so much and declare they don’t need it. I for one would love to try the SR1 if I could, before I conclude they’re not worth the money.
The equipment analogy was just an illustration.
How do you buy your wine? Tea? Coffee? With a blind test?
Personally I get the highest resolution content available because there is no cost, and then I play what I want through Roon without regard to resolution. A blind test has never occurred to me.
Come on now. Wine, coffee and tea varieties are like different artists, not like different formats. Do they sell 16/44 and 24/96 variants of the same exact wine? This kind of comparison only reveals how nonsensical is to charge differently for variants of what is supposed to be the exact same product. And what do you mean there is no cost? Hi-res tracks are 10-20% more expensive than CD. And if a blind test never occurred to you, why do you think you don’t need one? Don’t you want to know if you’re actually getting more for paying more?
I’m obviously talking about purchases, not streaming. That’s the subject of this thread. 10-20% is not minor. And I seriously doubt you wouldn’t care if you found out your RAAL-requisite was forged. Many forgeries can only be exposed with instrumentation. I know there are people for which money is no object, but that does not make scams right.
But we’ve drifted too much already from the subject. The answer to the question asked here is “yes”. I’ll leave it at that.
Yeah. In a way I feel sorry for the cynics among us.
But I think it depends on what music we listen to. I do almost all modern jazz and it all comes at 88 or 96 because that’s how it’s recorded. No scam.
I hear all kinds of horror stories about pop and rock, fake upsampling and different masterings and the volume wars. But I don’t do that.
I looked in my library at the few albums I have in 192k and they are all from the 60s or older, analog tapes, they are obviously not really high res but the people who believe those analog tapes are the greatest want to preserve the sound, ok fine, I don’t mind.