I’ve had a quick look and couldn’t see anything on the forum so far (which surprised me - happy to be pointed at appropriate posts), so here we go…
Basically, why buy hi-res downloads any more when Roon (or others) can up sample 16/44 for you? Particularly as some of the hi-res downloads are (I believe) not masters but taken from lower res originals. But even if they are from hi-res masters is there any difference in SQ?
(Why does that feel like I’ve just lit the blue touch paper and retired?!!)
Hi-Res, when done well, can be amazing (and, IMO, better than CD). This could simply be that we can avoid damaging reconstruction filters during playback, but perhaps not. However, I am continually surprised by occasional stunningly recorded/mixed/mastered CDs which wipe the floor with many nominally “Hi-Res” downloads.
Are these “poor” downloads up-sampled? Hard to tell without detailed analysis that is beyond most of us. Perhaps that’s why we need the “A” in MQA… [Fetches hat and legs it]
One of the best 192k downloads I have is the 2013 remaster of Kind Of Blue: really special, so perhaps something is captured in the encoding that matters?
In short, my experience is that Hi-Res has huge potential, but that potential is not realized if production is mediocre.
For those who, unlike you, don’t believe in audible differences between standard CD quality and high resolution audio, this might be interesting (perhaps life-changing ) news:
In 2016 Joshua Reiss, a Reader at the Queen Mary University of London (UK), published a paper titled “A Meta-Analysis of High Resolution Audio Perceptual Evaluation” in the renowned Journal of the Audio Engineering Society (Vol. 64, No. 6, June 2016). Here’s a short extract taken from the abstract of this paper:
“We undertook a systematic review and meta-analysis to assess the ability of test subjects to perceive a difference between high resolution and standard audio.” […] “Results 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.”
Of course, this doesn’t contradict your observation that
I also have lots of CD quality recordings (i.e. recorded, mixed and mastered at 16/44.1) that I think sound much better than many so-called “Hi-Res” productions. High sample rates are definitely not everything…
Hi @HWZ , That was a nice paper to read, thanks ! I think both you and the author are skipping to your conclusion a bit fast, however…
If I were asked to make an as-short-as-possible conclusion to his paper (way too short to be a scientific one…), it would look something like this :
Untrained listeners are incapable of discerning statistically relevant differences, between hi-res and redbook. Correct identifications are around 52%, which is almost the same as you you would expect from pure guessing (=50%).
Trained listeners seem to do a bit better : around 62% correct identifications.
The question you should ask yourself, is : how significant IS that 62% ?
How many correct identifications do you believe you’ll get, if you were tested for the difference between fresh vs. bottled orange juice ? (Because audible differences are often claimed to be THAT large…)
It is not very much more, than pure guessing. In other words : the difference is not very reliably detectable.
How meaningful is a difference, when you can barely detect it ? Wouldn’t you expect a near-100% rate, when differences are as audible as often claimed ?
Or in other words : in 38-48% percent of all cases, you think the ‘botlled’ juice is actually the ‘fresh’ one !
(You should know that I personally do not care for hi-res. I respect when people believe otherwise. But : please do not confuse your own experiences, with facts.)
Back on topic for @Phil_Wright : No, obviously upsampling is not the same, as recording an analog ‘something’ at higher sample rate.
Upsampling can never give you any meaningful frequency content above Fs/2. Actually recording the event at higher sample rate, can do that. (But the question remains : is that extra information really useful?)
Over sampling or a.k.a Up-sampling has been used in all CD players since Philips introduced 4x over-sampling in the early days of CD. So most of CD players today features 8x OS as a norm. In this case playing back CD 16 bit 44.1kHz is automatically over-sampling digital filter to 16 bit 352.8kHz.
The only difference is software up-sampling can have dedicated digital filters installed and tweaks applied to improve the transient response. For software up-sampling to be effective it must be set to maximum rate of the DAC. This will disengage it’s built-in OS digital filters. However, some specialised DAC can have its OS digital filter completely bypass.
Hi-Res on the other hand are mastered >16 bit and at least 88.2kHz and above. However, it will still need to OS inside the DAC to reach to its maximum sampling rate. Of course if the Hi-Res is 24/352.4kHz, then it simply pass through without OS!
Quite obviously, you got me wrong. I don’t think and didn’t write there’s a big difference in sound quality between redbook and Hi Res. On the contrary, what I did write was that I have lots of CDs that — IMO — sound much better than many nominally “Hi Res” albums. Nevertheless, the fact that Reiss and his team found out “test subjects can improve their ability to discriminate high resolution content” when “they receive extensive training” proves the difference under discussion isn’t inaudible. At the same time, Reiss’ paper also proves the difference is so small that many people (about 38%) don’t hear it (despite extensive training!). (BTW, I don’t really hear it, either. However, I’d never assume that just because I don’t, the difference doesn’t exist!!!)
I don‘t quite understand what you mean. Could you elaborate on this a little? In my experience, people usually have a hard time finding/agreeing on “the lowest common denominator“ in this context. Are you saying such findings aren‘t useful?
There are obvious reasons why Hi-Res will sounds better than Redbook, not so much on the details in the upper ends but Hi-Res mastering uses less steep digital filters on the A/D and D/A converters.
16 bit 44.1kHz though able to capture up to 22.5kHz, the steep cut off must happen before that to prevent aliasing. Pushing the sampling rate to 88.2kHz make thing better, no longer we need a steeper filter. The way this digital filters are designed greatly influence the SQ.
I just mean that, in this case, settling for what most people can or can’t hear won’t lead to a better understanding of the science, nor will it drive us forwards in terms of making real improvements.
Hi @HWZ ! I appreciate you correcting me ! I don’t think I got you wrong, however.
Of course a great recording on redbook, will be preferred over a lousy recording on hi-res.
No one would disagree. This has nothing to with the format itself being superior (or not).
I should probably have written “…ONE thinks the ‘botlled’ juice is actually the ‘fresh’ one !”, and "ONE should confuse his experiences…’
I did not mean you, specifically.
What I’m missing in the discussion, is : let’s say there indeed IS a small group of people, who can reliably detect the differences.
This paper does not provide any proof for that statement (unless I missed it), but sure, it’s a possibility.
The only thing I read, is : with training, a large group of trained and selected people have shown to be able to detect it, with a very high level of uncertainty.
I wonder aloud : Why do we care for differences SO small, that even selected/trained people cannot reliably detect them ?
There are far larger problems to solve, further up in the chain…
(Note : I’m not accusing anyone of being wrong, and neither do I disagree there is a difference.
But to me, it’s quite similar to polishing the paint on your Fiat Panda for less air friction/more speed. Yes, it can help. But you’ll have real trouble noticing it, and it is not very useful unless you have fixed the larger root problems, first)
Polishing your Fiat Panda - what a lovely analogy!
I suppose that gets to the crux of my question. Given a decent recording is it worth getting the hi-res version or would up sampling the 16/44 get as near to the SQ as realistically makes no difference? Gut feel says yes; it would be interesting to see if anyone has practical experience. The fact that I haven’t bothered trying yet probably tells you where my sympathies lie - too busy enjoying the music to spend time tinkering. Particularly as up sampling is free via Roon versus expensive downloads.
I don’t think my position is at odds with the polite roasting I got above - I totally agree that taking the argument to the extreme and up sampling low rate MP3s would sound awful. It’s more a question of where the law of diminishing returns kicks in.
Now as for MQA…grabs hat and legs it in same direction as Joel…
My opinion would be, well…, just an opinion. Wouldn’t be very useful for you, and you can probably already guess where I stand.
The only meaningful option would be, to form your own opinion on this. In a couple of hours, you will find all possible opinions below my post. Which of them is right for you ?
This points to a common difficulty across all of home audio recordings, digital downloads, CDs and also vinyl.
If you don’t know the providence of the recording (what was used for the master, how was it mixed or remixed, was the mastering engineer working for a great recording or just churning out material) you can’t really compare to another one release of a recording to another without taking into account the release. Kind of Blue is a perfect example, even on vinyl there are at least five mixes on the market: the original mono, the original stereo, the original stereo mixed down to mono, the speed corrected stereo and the third party LPs probably cut from one of the CD releases.
This is also the case with digital releases. How do you even start to compare file formats if you don’t know the origins of the releases?
I had absolutely no intention of “correcting” you. I really like your two witty analogies (orange juice and Fiat ) very much and I couldn’t agree more with your conclusion that
That’s why I explicitly stressed two things (with the help of Reiss’ paper, which is a meta-analysis of other studies):
a) The difference between CD quality and high resolution audio isn’t inaudible. If it were, listening training wouldn’t have the effect described by Reiss and his team (improved ability to discern the difference).
b) The difference is so small that many people don’t hear it (even if they receive extensive listening training).
Against this background, your question (why care - if the difference is so small?) makes perfect sense (IMO).
I can only answer your question from my personal point of view: Whenever an album I’m interested in “owning” is available in Hi Res, I buy the Hi Res version — not because I’m convinced I can hear the difference (in fact, from numerous informal single-blind ABX tests I’ve done, I know I can’t!). No, I buy it because music means a lot to me and I see no reason (except money) why I should settle for anything less than the (theoretically) best quality I can get. Fortunately, I’m lucky enough to be able to afford this.
) news: