Zp 24/96

Surely the point of hi-res isn't just the higher frequencies, but rather the perceived overall increase in 'resolution' achieved by the higher sampling rates?

You do not achieve "higher resolution" by using higher sampling rates. Since the Nyquist–Shannon sampling theorem definitively states that a sound wave can be perfectly reconstructed by sampling at a rate 2x the highest frequency, the only thing you achieve by increasing the sampling frequency is increasing the frequency range of the sound. Period.

Ignore the "jaggies" representation that you see in audio magazines, where the "jaggies" approach a smoother curve as you increase the sampling rate. This is a gross misrepresentation of the actual process. There is no "increase the samples and you get a better copy of the sound" process going on here. That is bunk invented by the ignorant and the snake oil peddlers. The actual science is quite different.

Jean_Lavallee,

I'll agree that the usual squared off diagram is not very helpful, but I'm not sure that showing the cells having rounded tops would be more intuitive for the lay public (or quite a few engineers) either.

Implied, but never stated, is that, after the high frequency reconstruction artifacts are removed, there is nothing left except the sampled fundamental.

---

I don't understand your circle analogy. Obviously the two points will define the minimum circle diameter, but there are quite a number of circles that can share that same chord.

You're right, I should have tought about it better.

Never been challenged yet, most people just draw using the points as diameter indication.

Thanks for setting me straight

You're right, I should have tought about it better.

Never been challenged yet, most people just draw using the points as diameter indication.

Thanks for setting me straight

Just change it to 3 points and the logic then applies.

A 44.1kHz sample rate can fully capture all of the frequencies in 20kHz+ audio signal.

OK - what about the 24 bit part?

It bears repeating that these guys are some of the leading experts in the world on this subject. THAT is why we all should be reading this article and taking it's claims seriously.

But then all you're doing is saying that 'my chosen expert is better than your chosen expert', where there are lots of 'audiophile experts' saying that the reverse is true.

With the quasi-religious undertones of the hi-res advocate, I can't see them being swayed just by pointing them at a single web-site - however highly you or I might rate them. After all, for all they know this web site could be to audio what Ken Rockwell's site is to photography.

I tend to prefer a scientific statement/theory/stance to be backed up by many others, hence my point about the single-source approach.

On source can easily be dismissed by the believer - dismissing many different reputable sources saying the same thing this is much, much harder.

amun,

In terms of rendering (playing into a room or headphone), controlled experiments have not been able to demonstrate any advantage for rendering more than 16 bits to a human listening panel.

I don't want to extrapolate this too far, but consider an interest calculation for a savings account. When the account is ultimately settled, the resolution of that payment is limited to the currency or coin available. One cannot pay fractions of a coin. And, what is the point of chopping a coin into a fraction? Would one need to purchase a fraction of a stick of gum? However, while calculating along the way, it is handy to use many decimal places in order to minimize errors caused by little truncations.

24 or 32 bits are very handy in the recording studio because they minimize issues caused by processing truncation, but we only need 16 bits for rendering. A prime benefit of recording at 24 bits is that the recording engineer does not need to "ride gain" (keep adjusting the recording level to match the artists). In order to minimize the impact of noise, one attempts to keep the levels high, but if the input signal level crosses over the maximum allowed, there will be clearly audible flaws in the recording. Recording at 24 bits relaxes this traditionally difficult aspect of recording.

When we "pay" our audio signal, we only need 16 bits.

---

If the end user wants to continue processing the recording, having access to the 24 bit data would be useful. But this is not the realm of the audiophile. It is interesting to note that I have never seen this mentioned in any of the heated discussions about 16 vs 24. All of the "heat" is directed toward somehow delivering the full 24 bits to the speakers -- without any processing.

---

BTW, while I can be a nightmare critic, picking up little clicks, pops, noises (in a local hall one really should close the men's room during a recording), and distortions, I'm not against constructive processing. For example, in a noisy listening situation a little dynamic range compression works wonders for listening enjoyment. And, the room could have an ugly frequency response issue that could be improved with a little equalization.

But then all you're doing is saying that 'my chosen expert is better than your chosen expert', where there are lots of 'audiophile experts' saying that the reverse is true.


Anyone who claims "expertise" in this are has to have a background in Audio Engineering or a related Engineering/Science based discipline. The Xiph guys have that in spades.

He is not the only one making these claims. His articles reference several expert sources that have conducted studies and research into this subject, including the Audio Engineering Society.

If you can find someone with an equivalent level of expertise in the subject who can point out specific flaws in the Xiph article based on Engineering, Science, or Mathematics then by all means present them. At the moment, until this is done, this article remains the nearest we have to the truth, the facts about better-than-full-resolution audio. And, as I said, Xiph are not some lone voice in the wilderness. Experts at the AES and places like HydrogenAudio and the Slimdevices forums have been saying this for years, but they have been drowned out by ignorance-based arguments from non-experts.

“There is a cult of ignorance in the United States, and there has always been. The strain of anti-intellectualism has been a constant thread winding its way through our political and cultural life, nurtured by the false notion that democracy means that "my ignorance is just as good as your knowledge.”
― Isaac Asimov

IMO that cult seems to have spread across the world.

I will point out that the word "audiophile" suggests someone who loves listening to audio. The people who call themselves "audiophiles" claim to be people who value the purity of audio reproduction. If this is the case, they should be welcoming this article as it points out a way to avoid introducing additional noise into their listening experience.

Cheers,

Keith

OK - what about the 24 bit part?

"Resolution" is a very misunderstood largely because of the pictures of jagged sine waves or analogies with photography and imaging that are so often used to explain it. These, usually, foster incorrect conclusions about how digital audio works.

When you increase "resolution", you are really just lowering the noise floor. An 8-bit sample will sound horrible because there is a lot of sampling noise. A 16-bit sample will sound better because there is less noise: the noise floor is lower.

Increasing the sample rate from 16 to 24 bits lowers the noise floor even further. However, a 16-bit sample rate is already capable of reproducing audio down to around -120dB, which is, for all practical purposes, beyond the ability of human hearing.

And, even in the quietest environment practical (like a commercial recording studio) the background noise will mask the quietest sounds, making the range of human hearing 100dB or less in practise. Even the most carefully mastered commercial audio material has a dynamic range less than 60dB which is well within the range of 16-bit audio.

As buzz mentions, 24-bit is useful during the music production stages where you may be mixing and manipulating dozens of tracks and applying processing (reverb, compression, etc.), but once the final product is completed, 16-bits is enough.

This is all explained, very well, in the Xiph article.

Have you actually read and understood it?

For reference, a lot of this stuff is also explained, to a much higher degree of detail and complexity in articles and books by Audio Engineers. One of the more famous ones is "Mastering Audio" by Bob Katz (ISBN: 0240 808376) who, by the way, is another expert in digital audio who should be taken notice of.

Cheers,

Keith

amun,

... If this is the case, they should be welcoming this article as it points out a way to avoid introducing additional noise into their listening experience.


Funds available for the audiophile hobby are always limited. Certainly, the available budget varies wildly between individuals, but there is always a limit.

An often asked question is "How much of my audio budget should I spend on [...]."

Here we are identifying an area where money is being wasted, unnecessarily trying to present 24 bits to the room. Channeling this same money into listening room improvements would have a much larger payoff.

Note that we are not suggesting that all DAC's sound the same, only that it is not the 24 bit part that one needs to worry about. Poor construction techniques can obliterate those last eight precious bits in a blink.