Vinyl v. digital curiosity

What I have seen, inverting one channel and sum to mono, on a mono record; will usually have the same amount of artifacts, never mind if the record is cut with a mono head, or a stereo head.

 

Re-worded

 

In the early days of CDs they were kind enough to warn us with those notes on the tray liner that the music is from analog sources and contains all sorts of defects and problems inherent to that domain. I was always so overjoyed to know that the history of great music was captured on crappy old fashioned tape and couldn't wait for the day when the labels would demand the sound be fixed and made better. Did they ever.

 

Actually, what I should have said is "Trade vintage and sealed! Be happy (and better off!)"

As I suggested in another thread response elsewhere, we probably have the greatest cache of sealed and/or Mint vinyl among our membership of any body on the planet -- we probably have the means among those n-100,000 members to resolve all of our wants-lists in an instant if we had an easier means of communicating what we have and what we'd take in exchange for it. There might be a way to do that P2P using the Discogs db. Will post soon...

 

Yes, that's exactly how I felt! I was soooo happy when CDs were introduced in the early 80's and I could afford a player. Loved them then and I still do 30+ years later. And all those original CDs I purchased back in 1984 - they still sound just as good as the day I got them. Can't beat that!

 

That's not "science" speaking but assumption based on appeal to authority (from my perspective as a practicing Prof. scientist of 30 years standing). What you describe here is the assertion that application of a theory, in this case Nyquist, will have perfect correspondence with reality if appropriately applied and measurably perfect. Unfortunately, most theories are never properly applied because the applicators are humans with mortgages to pay and reputations to protect. What usually happens is that underlying parameters are shifted until the results match with expectation (wherein the keen-eyed will see that all objectivity is gone), in this case "perfect sound" (forever). Yet, when I and many others (let us say 'analogue-philes') hear digital reproduction of higher-frequencies we hear something this is not merely imperfect, it is downright unpleasant to listen to to the point of being painful. I suggest to you that there is a a reason for that, and one that no amount of remastering, resampling, or increasing 'resolution' or sampling-rate will ever fix.

The Redbook standard for digital audio (16-bit/44.1 khz) provides ~ 44,000 samples/s for the ~ top of our hearing-range, ~ 20,000 Hz. That equates to ~ 2 digital samples for every vibration-cycle of the high-frequency harmonics that occupy this spectral-range. Two samples, one up and one down. That's Binary sampling. In effect, what A-D-encoding for CD does is reduce your 'transparent' medium from capturing audio at 44.1 khz to 0.0000441 khz at certain frequencies. Instruments of different types have different audio frequencies, and this is why they sound different, e.g., a kick-drum and a splash cymbal. Do you think that sampling a complexly-oscillating waveform at a particular frequency only twice (i.e., a square-wave) will capture all of the subtlety in that instrument's sound when that oscillation is 20,000 s-1? Do you think that the impoverished ability of CD-audio there will not be audible? In this vein, why do you think that everyone raves about digital bass and no-one writes home about digital treble? Moreover, do you think that increasing the sampling rate to 192 khz (or 384 khz, or whatever) will make any difference to the capture abilities in the audio band? Does 176.4 khz provide a four-fold increase in fidelity in the audio band, or merely the ability to capture a 88.2 khz audio signal (that we can't hear) with the same fidelity?

In terms of converters that "...measure properly", which one does? Which doesn't ring in the audio-band, which doesn't overshoot without its (presumably also transparent) brickwall filter? What aspect of Nyquist theory showed how pulse-code modulation captured the complexity of sound in both amplitude and frequency modulation? A sound is not just an on-off 'doink' -- it varies simultaneously in both Am and Fr domains, and will not be captured at high-frequency in anything like its entirety by simply sampling at 2*Fr.

If you like CD audio then you are fortunate, because not liking it is both a difficult and expensive place to be. I honestly could not be happier for you in that regard, but touting it as being 'measurably perfect' by bandying "science" around is not appropriate. Nature gave us ears to listen and assess with, and we all are perfectly equipped for that; assessing scientifically is far less straightforward and something we are not automatically equipped for. I have seen nothing in the Nyquist White Paper that describes non-frequency based fidelity to the audio captured, and I have heard nothing in digital audio (and I spent 30 years trying) that even approximates to the quality of analogue audio. As a realist and a parsimonious scientist, that is good enough for me.

Where it is perfect is in the copying, once digitised, and that is of interest to no-one but the archivist and file-sharer (and archival is uniformly agreed to be better catered for by pulse-width modulation which, perhaps unsurprisingly, is more 'analogue' in its digitisation).

 

The thing that no-one ever seems to pay attention to is how greatly our hearing deteriorates with age.

Having regular hearing tests due to my work there are lots of high frequencies i no longer detect. So I'd jokingly argue EVERYTHING sounds warmer to me now.

 

Digital often sounds unnatural when reproducing cymbals with a bright equalization. I suppose you just explained why.

 

I dont agreed i had heard, a mono record with a mono stylus and the record sound it more authentic sound but then is better to listen to it, whit one big speaker or headphones

 

Many early cd sound worse than 90's remasters

 

Warmer doesnt mean better!

 

LPCM is not less accurate at higher frequencies - the "only two samples per period” argument is a fallacy.

While it may seem counterintuitive, two samples per period is mathematically all you need to identify the original sine-wave. As the frequency of a sine-wave being encoded (note: any signal can be considered as comprising a set of superimposed sine-waves) goes up, the period (time for each cycle) goes down - once that period is equal to or smaller than the sampling interval the system breaks down and aliasing starts (hence the filtering before encoding to ensure nothing at or above the half-sampling frequency enters the system).

For 44.1 kHz PCM we are talking about content just under 22.05 kHz. No harmonics associated with a fundamental at that frequency will be captured but the system will capture a sine-wave just below the limit (not that anyone will hear it).

In reality, most 44.1 kHz sampled digital audio is fully rolled off by about 21 kHz.

Sampling and reconstructing an analogue waveform (encoding/decoding) is not a game of patching the original together from incomplete parts - it's a game of mathematical wizardry involving discrete, regularly spaced, instantaneous amplitude measurements (samples) and this sample-set's relationship to the original analogue waveform.

Mathematically speaking, if you take regular samples of an analogue waveform then, from these samples, you can perfectly reconstruct the original waveform up to the half-sampling frequency. There are no errors creeping in as the frequency rises.

In practice, the reconstruction is not perfect but not because of any frequency-dependent deficiency in the sampling theorem.

The biggest hurdle to transparency in digital audio systems is the filtering (before encoding and at decoding).

 

Only if it's poorly recorded or mastered. Otherwise, no issues.

 

Whoops - that should be:

"...once that period is equal to or smaller than twice the sampling interval the system breaks down..."

 

I have no idea what You mean, but if You say so,

 

I don´t think it explained anything, really. Maybe it explained that digital is complicated.

 

Let me rephrase that:

"...anything over two sample intervals per period is mathematically all you need to identify the original sine-wave..."


For clarity, the full edited paragraph:

"While it may seem counterintuitive, anything over two sample intervals per period is mathematically all you need to identify the original sine-wave. As the frequency of a sine-wave being encoded (note: any signal can be considered as comprising a set of superimposed sine-waves) goes up, the period (time for each cycle) goes down - once that period is equal to or smaller than twice the sampling interval the system breaks down and aliasing starts (hence the filtering before encoding to ensure nothing at or above the half-sampling frequency enters the system)."

 

But the number of sine-waves would be infinite if we're talking analog, wouldn't it?

 

Somehow this reminds me of the old riddle "how many grooves are there in an average LP?"

 

Yeah but it's not like the PCM system can only handle a certain number of sine-waves!

 

Hearing loss is not across the board. It varies for each individual. It has to do with genetics, a person's health, and a lot of it is due to environmental factors.

 

Because they are unique and often historically important. You can often buy a very good print for a lot less. Galleries often ban people from taking photographs of paintings as the flashes damage the paintings in the long term.

 

That's how no noise came about.

 

I like to hear the "brass" in the cymbals, and usually the analog playback is what can deliver this character of the instrument. Digital just sounds like a "sizzle". No character.

 

I think part of the problem of cymbals getting that sort of sound with digital is due to overuse of digital EQ and other digital processing during the recording and mastering. A good digital recording of an LP can capture a lot of that analog LP playback sound if you leave the digital side alone and don't mess with it. But if you start doing digital EQ and digital RIAA correction, and digital declicking and denoise and other digital processing to fix the needledrop you begin to lose the analog sound and end up more digital. I've heard some impressive digital needledrops that did no digital processing or fixing. I've heard some unimpressive and digital sounding needledrops that did a lot of digital processing to clean up the sound. The ones with no digital cleanup or digital post processing often sound better.