Someone who knows more than I do should chime in here, but I imagine there will be stuff above 22kHz in the hi res version that is not there in the redbook one?? So doing a difference signal should show that if nothing else...
What sounds are being filtered in an upconverted from CD res audio file? And have you verified this in an A/B listening test? What do you listen for that indicates this is happening?
I think a lot of the confusion around this question can be cleared up by referencing some basics of digital sampling theory. For example: Any sound frequency can be digitally sampled and then accurately reproduced if it is sampled twice. So the necessary sample rate is twice the highest frequency you want to be able to encode and reproduce. For humans that's 20kHz, so you need a sample rate of 40kHz - plus some cushion for digital reconstruction filtering. For that reason, 44.1lHz is sufficient for representing all sounds that human beings can hear. It is simpler - and in the past was easier and cheaper - to create a reconstruction filter if you have more cushion to work with, and so for that reason some folks prefer their digital music to use a 48khz (or higher) sample rate, because you never know how gradual/lame/half-a**ed the filter design of your playback gear might be. But with that said, today it is trivially easy and inexpensive to engineer and implement a reconstruction filter that can do its job in the cushion area between 20kHz and 22.05kHz (the latter being the highest frequency encodable with a 44.1kHz sample rate). And many cheap DACs do just that. Higher sample rates produce exactly zero additional "resolution," "detail," or "refinement" in the audible hearing range. This is not a perceptual argument I am making. Rather, it is a mathematical fact. Higher sample rates can only encode higher frequencies. This makes intuitive sense if you think about the fact that a bass drum at 100Hz is sampled 10 times more than a vocal at 1kHz, and 100 times more than a cymbal at 10kHz. This is true no matter the sample rate. So do we all feel that in every single one of our digital music recordings, the bass drum sounds more "high res" or "refined" than the vocal or the cymbals? Of course we don't. Putting aside the reconstruction filtering for the moment, a 100Hz sound can be just as accurately recorded and played back with a 200Hz sample rate as with a 44.1kHz or 96kHz sample rate. That's just a fact. That leaves bit depth. Here again, increased bit depth does not increase "resolution" or "refinement." One PCM bit = 6.02dB, no matter how many bits you use. A 24-bit recording will allow for greater dynamic range than a 16-bit recording, to be sure - but all that extra dynamic range will be at volume levels below the minimum volume level/noise floor of 16 bit. So a 16 bit recording will have a noise floor of about -96dB. If you listen with magical noise-cancelling headphones that block literally 100% of the ambient noise in your listening space so your background is literally zero dB total silence, and you listen to music on those headphones at dangerous levels above 96dB, then in the short period of time before you damaged your hearing you could potentially hear the noise floor of the 16 bit digital format during moments of silence between tracks. Even in this nonsense scenario, we'd have to ignore the sound of our own breathing, which is at a volume level of about 10dB. And the 16 bit source would have to have no dithering applied - because of how human hearing works, dither is easily able to bring the effective noise floor of 16 bit sources down to about -120dB. That's below the threshold of human hearing in any circumstances. I agree 100% with folk who talk about the value of oversampling and high bit depths during recording, processing, volume alterations, and so on. But for the end product that we listen to, 16/44.1kHz is absolutely, positively sufficient - and more to the point, higher sample rates and bit depths are simply incapable of doing most of what is claimed for them in terms of improved "resolution." I think folks might be surprised if they used a good quality sound pressure level meter and measured (a) the ambient noise level in their listening space, and (b) the hiss/self-noise of their gear as it comes out of their speaker's tweeter. The result of those two measurements will usually reveal that the noise floor of one's room and gear is much higher than the -120dB that dithered 16-bit provides - and for that matter higher than the -96dB that undithered 16 bit provides. Heck, for that matter, just find a test tone online, burn it to a CD-R, and calibrate your volume level so the CD-R is playing at -96dB. You'll never be able to hear it in any home listening space.
I boosted the treble/kick drum in Audacity at 32bit Floating point on the Waterwings track off the Lee Ritenour Friendship album CD. It's a very well recorded album with a lot of high frequency instruments with rather noticeable reverb tails and decay all sounding better than the vinyl I had back when it was first released in 1979 after Waterwings was presented on the Johnny Carson show back then. The vocal that starts out in unison with the electric bass is a bit obscured. I don't hear any bad sounds that need to be filtered even on the YouTube video so I don't know how high resolution is going to make this track sound better. But I was able to bring out the vocal bass intro to hear sibilants. I was also able to reduce gain in the soprano sax and bring out the sound of the reed.
The playback sampling rate is moved higher. Instead if 44.4kHz, it is usually 96kHz or 192kHz. This moves the output reconstruction filter corner frequency in the DAC correspondingly higher and this can make a difference in the audio band. This has nothing to do with the original sampling rate of the studio master, but ideally a high resolution master is distributed as a file with the same sampling rate. I don't want to get into an ASR-style pissing debate with blind tests and all that A/B nonsense. There are real technical benefits in having a higher sampling rate which go beyond simple high frequency extension (although this is also a benefit, but a questionable one). The bit depth is also usually higher which gives greater signal to noise ratio, but this is not as much a benefit in my opinion. In the end, you have the choice to listen to whatever format you wish and which provides value to you.
About 3 years ago I recorded an acoustic guitar track using Audacity on the default setting (32 bit float, 44.1 khz). I then exported a 16 bit and a 24 bit version. I later decided to give these two versions a comparative listening. I was not expecting to hear much difference, thinking that 16 bit was "good enough." My ears quickly told me something else. The 24 bit version sounded substantially better. I was listening with a pair of entry-level Grado headphones (SR 60s) plugged into a Dell laptop, not on high end audio equipment. That settled the matter for me: hi res sounds better than CD quality, at least to my ears. I'm pretty sure my hearing was average for a 50 year old dude. However, I had learned to be a careful listener as a result of my audiophile hobby. Can the general population discern the difference between hi res and CD quality? I don't know and, frankly, I don't care. I purchase music for my own enjoyment. I always get a hi res version when it's available.
This is only true for signals that can be decomposed into sinewaves within the Nyquist limit. I assumed everyone on this thread was already familiar with Shannon's sampling theorem. If not, here is his paper, it is a good read: Wayback Machine Since that is already well known, that is why we have been focusing on timing of time-varying signals (the start and stop of which cause a spectrum that does not fit within the Nyquist limit), inter-modulation distortion from harmonics, etc.
You may need to start from scratch, existing measuring methods fail to correlate with SQ differences.
I give it a B, overall. More consistent than the vinyl LPs in my collection, which can range from a D to an A-. When comparing formats of the same record, sometimes the CD version is a C, while the LP is an A-. More rarely, the LP version is a C+, and the CD is a B or B+.
I bought my first CD player in 1985 and enjoy the format even more today. The quality of sound depends on many other parameters, but resolution is not essential at all.
There's one other slight condition. Shannon (and Nyquist) mathematics are explicitly referencing, and depend upon, infinite duration signals. As humans we will never realize that. Our various engineering approximations can work well, but they are engineering approximations. Throwing higher levels of engineering at an approximation will make it a more faithful approximation...though the implicit point of this thread is, "What level of approximation will allow me to fully appreciate music?" What I find especially enlightening are the first two paragraphs of Shannon's chapter XII Continuous Sources. We are talking about musical signals after all. Here's Shannon's text starting XII Continuous Sources: "If the source is producing a continuous function of time, then without further data we must ascribe it an infinite rate of generating information. In fact, merely to specify exactly one quantity which has a continuous range of possibilities requires an infinite number of binary digits. We cannot send continuous information exactly over a channel of finite capacity. "Fortunately, we do not need to send continuous messages exactly. A certain amount of discrepancy between the original and the recovered messages can always be tolerated. If a certain tolerance is allowed, then a definite finite rate in binary digits per second can be assigned to a continuous source. It must be remembered that this rate depends on the nature and magnitude of the allowed error between original and final messages. The rate may be described as the rate of generating information relative to the criterion of fidelity." So it comes down to, on an individual basis, what degree of engineering approximation are we happy with?
My initial reaction to this is that your interpretation or application of the underlying principle is not correct - but I want to reserve judgment, so let me ask: what exactly are you claiming is audible sound that cannot be captured with a 40+ kHz sample rate? Are you saying that some sounds in recordings are in the audible range but cannot be decomposed into sinewaves within the Nyquist limit? Are you saying that 44.1kHz is not "fast" enough to encode some sounds in recordings?
I agree that the benefits of a higher sample rate are indeed as you say - which is why virtually every DAC oversamples the input signal internally before spitting out the reconstructed analogue signal to the output stage. But that is not the same thing as saying that the input - the digital music file or disc - needs to have a sample rate higher than 44.1kHz or that there are sonic benefits to feeding an oversampling DAC with a 96k signal instead of a 44.1k one. I also agree that higher bit depth provides a lower noise floor, which is very helpful during digital mixing, processing, and production - and can also be helpful in one's digital playback chain if one uses a lot of DSP (digital signal processing) - for example room-correcting EQ. But for a typical stereo playback use-case, I don't see the benefit or need for bit depths in excess of 16 bit when it comes to the digital music file or disc. The very best DACs are capable of about 21 bits' worth of S/N ratio, and the very best power amplifiers are capable of about 19 bits' worth - and that's at very high volume. At normal loud listening levels, in a quiet listening space with ambient noise around 25-30dB (and that is indeed extremely quiet for a typical home), the perceivable noise floor is not going to be anywhere near 96dB. Again, no disagreement that higher bit depths lower the noise floor, and no disagreement that higher bit depths and sample rates are useful for recording and production. And no disagreement that oversampling is beneficial when it comes to the DAC stage.
Maybe the intermodulation distortion that hi res files than impart on amplifiers ,tweeters and headphones in particular is what is causing the slight differences you are hearing? As we know from tube amplifiers 'distortion' in the chain can artificially heighten the sense of timbre/soundstage .Sometimes to a greater extent than is on the actual recording.
A question for those who think 16/44.1 is not enough: How much you would state is definitely sufficient? 24/96 is enough or not? 24/192 is enough or not? 32/384? Where do you put the limit? How much is too much?
typically they don't use a constant over-sampling rate. My DAC: 128x 44.1 64x 92 32x 192 The filter can be optimized for a constant rate. Only the cutoff needs adjusted.
Isn't the industry standard for recording 20/48? You can't extract any more information than the source provides. When discussing generating making files from analog tapes the source is the limit and 16/44 is more than sufficient. imo opinion regardless of the rate you will lose something.
The "Geniuses" y0u mention, had profit and copy protection in mind first. There is a detailed essay about why SACD was created, and it had nothing to do with better sound, but "Selling" audiophile on the promise of a new format that was supposedly better. The reality, it never delivered at all.
Feeling like Nyquist and Shannon would be great names for pets (cats). Not taking anything away from the very successful thoroughbred race horse named Nyquist. one of whose foals was aptly named Shannon.
I think the deeper issue, is that most are not able to carry through with your certainty, and ability to discern this when put to a test. If it was as certain as YOU claim, then show me a test study of 25 guys just like you that show ability for beyond guessing that can tell Hi-res from CD quality sound. No one it saying what YOU are hearing, but if you make a big claim, there should be some burnen of proof or at the least an ability to demonstrate it.
