First, many important instruments radiate sound energy far above 22.05 KHz, the theoretical bandwidth limit of 16-bit 44.1 KHz sampled audio, and while this "ultrasonic" energy may not be directly audible, it significantly affects listener's subjective enjoyment of recorded music.

It's Alive! Ultrasonic Spectra Isn't So Ultra Anymore

Second, properly set up vinyl systems can reproduce high order harmonics well above 40 KHz.

Analysis of Vinyl Frequency Content

Doesn't vinyl have a theoretical bandwidth limit too? How do we test this limit? Is vinyl's reproduction of high frequency signals limited to harmonic overtones?

For now I've lost all fondness for digital audio, at least until 96 or 192 KHz sampled music becomes easily accessible as 16/44 is now.

I ran the numbers a while ago assuming a fixed 70db SNR and a 20khz bandwidth for vinyl, and CD came on top by a smidge. If you bumped vinyl's bandwidth up to 100khz then vinyl would win, but really - at any frequency over perhaps 10khz, it's very unclear if one really has 70db of SNR to work with. Mistracking (or physical destruction of the cutting head!) becomes a much greater concern at loud high frequencies.

None of this really matters though, because nobody has ever shown with certainty that ultrasonics are audible. Oohashi's paper is a joke (search HA for more details). Meyer/Moran is a far, far, far better test. They couldn't find any evidence of audibility.

Quote:



That theoretical 44.1KHz bandwidth limit is just that. It was devised when digital to analogue converters were of the non-oversampling types. Overpsampling DAC chips don't have that 44.1KHz limitation. Some (many) DAC designers are still fixated on the 44.1KHz cut off for their anti-aliasing filter, but the ones who have seen the light and the advantages of oversampling are now taking digital audio to new levels of enjoyment. And I say this as someone in the inside of this process.

Quote:



Oh yeah? When I started my life into audio engineering all of us at uni had to sit a hearing and eye test. And there were people on my course who could hear tones up to 23KHz. That type of test is still done today in all sort of medical and professional circumstances.
But leaving those tests aside, my neighbour's dog can disprove with certainty that ultrasonics are audible. There is also an anti-hooligan device out there that relies on emitting an ultrasonic signal of deafening proportion, which is only audible to young people.

Quote:



I think the people who make the source components just follow what the most popular format is. If a CD is properly mastered it should be bandlimited to 44.1/2 anyway to avoid aliasing, and an oversampling filter can't bring back high frequency content that was filtered out in the recording stage. SACD looks promising, but simply doesn't have as many releases as vinyl.

Quote:



Sadly no it does not. The Oohashi paper is seriously flawed and their effect was recreated and shown to be straightfoward IMD not an effect of supersonic frequencies.


Quote:



You use a first year undergraduate paper as a supporting argument ? Sadly while this is a farly typical ugrad paper it is hardly the stuff of scientific rigour in fact this is just a slighty sub par paper for someone in a purportedly scientific discipline riddled with slightly emotive and unsupported assertions, very poor references and calls to non academic sources. This is a somewhat biased paper and ignores a great deal of recent research in the field.

Quote:



Also AES papers back in the late 1970s showed just how irrelevant high frequency (20K+) content was when it came to listening to music. When they attached filters at 16K and 18K (let alone 20K) to music with high frequency content the effect was not noticeable.


Quote:



Fine, but unless your listening habits includes chilling out to 23K test tones it really just makes no practical difference when the high frequency content is embedded in music where the lower frequencies ( < 10k) tend to dominate to a massive degree. Yes you could deliberately record music with massive amounts of energy in the upper ranges, but why would you want to ?

Quote:



This really is not relevant for listening to music

Could you provide some references to the sources you are talking about so we can learn more about this?

Quote:



Sure.

[1]The effects of low pass flters
-------------------------------

Sampling-Frequency Considerations in Digital Audio
TERUO MURAOKA, YOSHlHlKO YAMADA, AND MASAMI YAMAZAKI
JAES (1978)

This paper describes the effect of filters at 14, 16, 18 and 20K. I will quote the conclusions verbatim

1) 20-kHz cutoff is considered to be quite sufficient,
without deteriorating sound quality (including the effect
caused by the phase characteristics of low-pass filters).
2) The critical cut-off frequency is considered to be
about 15 kHz.
3) In designing digital audio systems, the sampling
frequency should preferably be set to cover the 20-kHz
bandwidth. The sampling frequency of 44.05594 kHz
satisfies this requirement.(*)

* - At this time JVC had their own digital sampling ideas


[2] Replication study on the Oohashi paper
-----------------------------------------

Detection threshold for tones above 22 kHz
Kiryu, Shogo; Ashihara, Kaoru (**)
AES Convention (2001)

A verbatim quote from the discussion

The subjects could discriminate the stimuli with and without
ultrasounds only when all components were mixed and presented
through the same loudspeaker. When as many as 6 loudspeakers
were used in order to prevent intermodulation, no one could
discriminate the stimuli.

** - did an interesting paper on jitter as well.


[3] On the naturalness of SACD vs DVD-A and inherent faults of PCM
----------------------------------------------------------

DVD-Audio versus SACD Perceptual Discrimination of Digital Audio Coding Formats Listening Comparison Test between DSD and High Resolution PCM (24-bit / 176.4 kHz)
Dominik Blech and Min-Chi Yang (2004)

Verbatim quote

Of a total 2,900 choices (145 test sequences × 20 choices per test sequence) there were 1,454 correct choices and 1,446 incorrect ones (see Figure 10). This result comes remarkably close to that which would be
expected (arithmetic mean value of 1,450 correct and 1,450 incorrect responses) in a statistically “pure chance” experiment. The four extra correct choices (not to be confused with the four test subjects who attained
critical probability with their test scores) represents a deviation of only 0.28%. Even with signals that had very short rise times (percussion and harpsichord), the digital encoding methods of the sources could not be
distinguished from one another

cool! I'll have to look those up myself

There was a study where recorded music was played with everything above 10kHz rolled off. People were asked to compare it to full range recordings and say which one was better. The vast majority of people said that there was no difference.

The frequencies that are important to recorded music are between 40Hz and 10kHz. 20-20 provides a handle a full octave above and below that. The balance of the response between 40 and 10 is MUCH more important than extension beyond that.

Frequencies beyond the range of human hearing, although perceivable to some degree, make absolutely no difference to the enjoyment of recorded music.

See ya
Steve

Quote:



Vinyl doesn't go through digital conversion of any form. The less processing the signal goes through from the microphones, the better it is. Live music or being there is as good as it gets. Anything added or subtracted from that point on just does damage.

Personally I love the sound of DVD-A. I enjoy listening to my audio DVDs more than any CD I own. I have just started to get into vinyl a little bit, but so far I am unimpressed (it is still early to make a judgment).

Quote:



If you think that vinyl represents a purer path from microphone to ears then I am afraid you are seriously mistaken. Ask a recording engineer about the steps required to get a studio recording from master tapes ready for a lacquer then look into the process of getting an actual record pressed from a lacquer. Then how about the RIAA curve - a massive piece of processing.

Prefer vinyl all you like but please do not mistake it as a pure path.

Quote:



Well, that is only true for the first stage of recording. When you record something, you inevitably lose something. Beyond that, it takes well judged adjustments to adapt the sound to suit the medium. Those choices improve the sound. The goal of recording isn't to recreate a live performance as accurately as possible. It's to create a performance that goes beyond reality, creating its own presentation. It's like filmmaking... you can make a drama or a documentary, but neither is the same as reality. They are manufactured realities that transcend realism.

Nick is absolutely right. There are lots of compromises and adjustments to be made to a recording to make it work well on vinyl. Mastering is a much more complex process for vinyl than it is for digital.

See ya
Steve

Quote:



I fully realize that the processing path to a finished LP is flawed. Does the 2 track master suffer more damage when processed into an LP or a CD? This is the real question.