[audiobat]

Clarification:
I bet that the average audiophile only plays 15 to 20 percent of their music collection, ( I should have added) on a regular basis.

Help me if I am off base with any of these calculations (no proctologists please).
I realize some people have various sized music collections, so results may vary.

1000 cd's or LP's X 15% = 150
1000 CD's or LP's X 20% = 200
55 minutes per cd or LP = 137.5 to 181 hours

Free time per month:
30 days x 24 hours = 720 hours
sleep 6 hours a day = subtract 180 hours
Work 40hrs x 4.33 weeks = subtract 173.2 hours
366.8 free hours per month not working or sleeping

137.5 to 181 hours = 15% to 20% of a 1000 cd or LP collection.

Listening to just 15% to 20% of that 1000 CD or LP collection would take up 37% to 49% of the available free time per month.

Humm...

[jimdgoulding]

I've had some of my records since I was at least 18 and that puts them at 40+ years old. I'm pretty much convinced that records are actually pretty robust considering everything, and the only caveat is that they always seem to have small "accidents" happen to them.

With a properly tracking, lower VTF cartridge, proper anti-skate and VTA, there shouldn't be any reason why you can't get hundreds of plays without detectable wear. I should also add that good cleaning habits would play a large role in record life.

W

+1

[audiobat]

With regular cleaning and sensible care an LP will last as long as one could want.
Maybe the wear issue is more a factor on a phono cartridge?
I have been through more worn out cartridges than worn out LP's.

[dlparker]

Uh, Mr. Wayner, there is a group of German gentlemen wearing powdered wigs outside who wish to discuss your comment with you...

D.D.

And let's not forget Uugah's Concerto for 2 rocks and a Mastadon Tusk.

[Wayner]

What I'm curious to find out, considering how much many of us buy used records, is how does a record hold up when these criteria are not met...

I'm guessing that 98% of the record buying public did not have a properly tracking, etc etc setup.

We will get to the bottom of this.

I have bought my share of used LPs, just cause I could not find it remastered. I do pull the LP out of it's sleeve and look at both sides, looking for scratches, and wear marks in the groove, that sometimes take on a dirty smoky look. Sometimes that is just plain dirt and gook in those grooves.

I bought an Aaron Copland LP a few years ago that I thought was totally garbage popcorn. But I cleaned it a few times, played it a few times and it literally cleaned itself up.

Jeff, if we knew the answers to this we'd be even more millionairs 

Wayner

[TONEPUB]

Clarification:
I bet that the average audiophile only plays 15 to 20 percent of their music collection, ( I should have added) on a regular basis.

Help me if I am off base with any of these calculations (no proctologists please).
I realize some people have various sized music collections, so results may vary.

1000 cd's or LP's X 15% = 150
1000 CD's or LP's X 20% = 200
55 minutes per cd or LP = 137.5 to 181 hours

Free time per month:
30 days x 24 hours = 720 hours
sleep 6 hours a day = subtract 180 hours
Work 40hrs x 4.33 weeks = subtract 173.2 hours
366.8 free hours per month not working or sleeping

137.5 to 181 hours = 15% to 20% of a 1000 cd or LP collection.

Listening to just 15% to 20% of that 1000 CD or LP collection would take up 37% to 49% of the available free time per month.

Humm...

Agreed, but I listen to music 12-16 hours a day, sometimes longer...

[Freo-1]

This thread is going off the rails a bit. 

The major issue of this thread is: Is High Resolution Audio Better (than CD?). 

I think the short answer is: Yes!

Not only does high resolution digital have better frequency extension and noise floor, the data reconstruction of the digital data to a analog waveform is superior from a number of technical standpoints (Shannon's therom does not (directly) play into the reconstruction portion of the waveform).

The best recoding medium available right now is the high resolution digital recorder and the best overall playback media overall is a high resolution digital playback device.  From an engineering (or performance) standpoint, this is not even up for debate.

High resolution digital also provides multi channel recordings, something analog (vinyl) does not support. (less the 70's quad).

The discussion about vinyl is OK, but not germane to this thread.  From an engineering standpoint, vinyl will always have more technical limitations than high resolution digital.

[trebejo]

The crinkly stuff on the waveform that you are trying to reproduce will not be reproduced by your ears. That's the problem.

If you believe that your ears can benefit from additional information above 20khz, that's a different matter. I certainly don't, and apparently, neither do audiologists.

However if you have more information from folks that study the innards of human ears, we're all ears. As far as we know, 20khz is it.

btw the previous stuff you posted was basically all about the production side of things, sound effects, etc. Once the bits are ready to be put on the CD, then 44khz is all you need.

Nothing's changed, theorem still stands. Sorry.

[Freo-1]

The theorem IS NOT directly related to the reconstruction of the analog waveform between 20Hz and 20 KHz, and THAT is one of the primary reasons Hi Resolution sounds better. 

Your argument continues to overlook the density of the bits regarding the analog waveform.  The theorem, as stated earlier, can apply, and High Resolution will still sound better, due to more precise reconstruction. 

Recommend you study up on the digital to analog conversion more, and pay attention to ALL the filtering aspects, and why the denser bit pattern helps with resolution.

Please read the following link, and it clearly explains WHY it is better. 

http://patches.sonic.com/pdf/white-papers/wp_dvd_audio.pdf


Many of the extra bits are within 20 to 20 Khz, so what part of a better analog waveform reconstruction are you failing to grasp?

"  Sample                         Stereo"
  Size     Sampling Rate       Transfer
 (Bits)  Samples Per Second      Rate
16     44,056 (44.1 kHz)**  1409 kbps
16     48,000 (48 kHz)      1536 kbps
24     88,200 (88.2 kHz)    4234 kbps
24     96,000 (96 kHz)      4608 kbps
24    176,400 (176.4 kHz)   8467 kbps
24    192,000 (192 kHz)     9216 kbps
24    384,000 (384 kHz)    18432 kbps
Compressed MP3 files (for comparison)
Common MP3 file:    128 kbps
High-res MP3 file:  320 kbps
** Standard music CD (see CD-DA)

[trebejo]

Ok, I took the time to download the 16-page pdf and I don't see what you are talking about. Page number?

[Freo-1]

Kindly look at page 5 to see an example of the bit density deltas.  THAT is where the high resolution is better to reconstruct the analog waveform.  One gets better resolution, hence better fidelity within the audio spectrum. 

If you have ever worked with analog synchro circuits, it's like the difference between a single speed (1X loop) and a two speed (1X and 36X) loop, where more precise data can be processed.

[trebejo]

Ok, yes, once again, that is like the previous graph from the Indiana University website.

btw I neglected to reply to that earlier post but that is a very interesting paper over at Indiana (one of the better music schools afaik).

Now the graph there shows what's going on pretty clearly. You get a crinkly "real" wave, and then due to interpolation you lose some of that crinkliness. However, such a loss is associated with frequencies that rise above (well above, even) than 20khz.

A step wave, just to cite a rather famous example, uses an infinite spectrum.

So so far, all in agreement. The disagreement begins with what takes place inside the human ear. Unfortunately, to our "impaired" ears, all the information above 20 khz is superfluous. Our ears will not be able to distinguish between the crinkly wave and the less crinkly wave. That's what the audiologists tell us, at least; that our ears are only good out to 20 khz.

Now if you're doing some signal processing when you are recording, then that's a different matter. It may very well be the case that the recording process, and obviously the editing process where sound effects are added, all benefit from a higher sampling rate. That is the way that I interpreted the IU web page.

However once you have the DAC waiting for a string of bits to transform into a 20 hz -- 20 khz sound pattern, then you will not need more than a 44 khz sampling rate to feed into the DAC.

There may be some confusion in the term "sampling rate", and maybe I'm not using the correct term here because I do NOT mean that the original microphone, etc. are not going to do a better job if they sample above 44 khz. What I mean by "sampling rate" is what the DAC will see when it will convert it into an analog waveform which is ready for the preamp.

There was a post earlier where someone else doubted that you could properly reconstruct a 10 khz waveform when only about four samples of the wave were taken. He was being charitable, too! He could have complained about resolving a 20 khz wave with a mere two points on one of its cycles... well, that's why the theorem is so cool, it shows how to do it. It's like a whodunit--who made that wave? By a process of elimination, all the coefficients are determined because they are the only combination of those coefficients that could produce that particular pattern of 0's and 1's, when sampled up to twice the given frequency range.

Shannon's theorem does not state that you don't lose information; it only states that you don't lose information up to half the sampling rate. With the recording process, and in particular with all the filters that are applied (which should escape us if we're not in that biz), there may very well be benefits to upsampling. But to the reproduction of information from 0-N hz, you should not need more than 2N hz because what you will lose will--according to the theorem--be above N hz.

I must politely ask now, have you worked through the theorem? Do you know why it works?

[Chromisdesigns]

The theorem IS NOT directly related to the reconstruction of the analog waveform between 20Hz and 20 KHz, and THAT is one of the primary reasons Hi Resolution sounds better. 

Your argument continues to overlook the density of the bits regarding the analog waveform.  The theorem, as stated earlier, can apply, and High Resolution will still sound better, due to more precise reconstruction. 

Recommend you study up on the digital to analog conversion more, and pay attention to ALL the filtering aspects, and why the denser bit pattern helps with resolution.

Please read the following link, and it clearly explains WHY it is better. 

I don't think so -- first, that paper contains the following statement:

"When digitizing an analog signal, better resolution (higher fidelity) may be achieved by
increasing the sample-rate (number of samples per unit of time), the word-length
(number of bits per sample), or both."

THAT is simply untrue, per the Nyquist-Shannon Theorem.  Any sampling rate greater than 2x the maximum frequecny being sampled encodes the original signal 100%.  That is the essence of the theorem, and the basis for digital signal processing, including audio analog to digital conversion.  Oversampling provides no additional information in the digital bitstream.  None.

Now, if you go on to look at the other side of the coin, the digital to analog conversion, the Theorem still holds, but with a couple caveats:

1st, N-S theorem as it applies to reconstruction of the analog signal from a bitstream involves summing a series of sinc functions.  Most (all?) DACs use square waves, though.

2nd, 100% reconstruction of the original signal is only guaranteed when the sinc functions are summed over infinte time, as each sample contains information about the entire waveform.  Real devices, of course, can't do this.

So, you might think that these two caveats would imply that more bits equal better resolution, at least in the DAC side of the process -- except that the error terms due to both of these issues are well beyond the audio spectrum!  In other words, you can't hear them.

A finite series of square waves can approximate ANY sine function (and thus sinc functions).  I believe it's been shown that as long as sine waves up to at least the 5th harmonic of the fundamental frequency are included in the series, any error terms again exceed the audible bandwidth.

There are similar results for the required amount of time over which summing occurs in the DAC.

[doug s.]

it has been well documented that, a)-ultrasonic sounds affect the sounds below 20khz.  and, b)-studies, with folks who can hear, and those who are deaf, indicate that people can perceive ultrasonics even if they cannot "hear" them.  and, it has also been documented that normal musical instruments emit ultrasonics in warying degrees.  perhaps this is why some folks perceive better sound when "supertweeters" are introduced in some playback systems, even tho the sounds above 20khz cannot be "heard". 

some interesting studies (there are a lot of studies on the 'net):
http://www.cco.caltech.edu/~boyk/spectra/spectra.htm
http://www.tinnitus.vcu.edu/Pages/Ultrasonic%20Hearing.pdf
http://www.sciencemag.org/content/253/5015/82.abstract

doug s.

[rbbert]

And if 16 bits is good enough for "perfect" reconstruction, why not 14 bits, or even 12?  12 bits still gives us 72 dB signal above noise, about like a semipro open reel deck at 15 ips.

[Chromisdesigns]

it has been well documented that, a)-ultrasonic sounds affect the sounds below 20khz.  and, b)-studies, with folks who can hear, and those who are deaf, indicate that people can perceive ultrasonics even if they cannot "hear" them.  and, it has also been documented that normal musical instruments emit ultrasonics in warying degrees.  perhaps this is why some folks perceive better sound when "supertweeters" are introduced in some playback systems, even tho the sounds above 20khz cannot be "heard". 

some interesting studies (there are a lot of studies on the 'net):
http://www.cco.caltech.edu/~boyk/spectra/spectra.htm
http://www.tinnitus.vcu.edu/Pages/Ultrasonic%20Hearing.pdf
http://www.sciencemag.org/content/253/5015/82.abstract

doug s.

OK, but the question arises how an ultrasonic sound can either "affect" those below 20 khz, and how someone could perceive them (through hearing, that is...I can certainly see how ultrasonics might be otherwise perceived, especially at high energy levels).

As far as I am aware, there are only two ways -- via subharmonics, and via intermodulation distortion products with a result in the audio range.  So, yes, I'd agree that "effects" of ultrasonic sounds might be heard (though not the fundamental frequencies, by definition).

However, those "effects" are ALREADY PART OF THE SOUNDFIELD when recording music!  So a well-made recording has those effects in i,  within the audible frequency range, does it not?

And, so, no oversampling should be required to digitize them.

[Chromisdesigns]

sorry, post got duplicated.

[trebejo]

In true audiophile fashion, we all have our own truths. 

[Chromisdesigns]

As a side note, there have been successful attempts to design audio transducers that produce audible sounds via differencing two ultrasonic beams.  Which is one example of being able to hear **effects** of ultrasonics, but not the base tones themselves.

[Freo-1]

I don't think so -- first, that paper contains the following statement:

"When digitizing an analog signal, better resolution (higher fidelity) may be achieved by
increasing the sample-rate (number of samples per unit of time), the word-length
(number of bits per sample), or both."

THAT is simply untrue, per the Nyquist-Shannon Theorem.  Any sampling rate greater than 2x the maximum frequecny being sampled encodes the original signal 100%.  That is the essence of the theorem, and the basis for digital signal processing, including audio analog to digital conversion.  Oversampling provides no additional information in the digital bitstream.  None.

Now, if you go on to look at the other side of the coin, the digital to analog conversion, the Theorem still holds, but with a couple caveats:

1st, N-S theorem as it applies to reconstruction of the analog signal from a bitstream involves summing a series of sinc functions.  Most (all?) DACs use square waves, though.

2nd, 100% reconstruction of the original signal is only guaranteed when the sinc functions are summed over infinte time, as each sample contains information about the entire waveform.  Real devices, of course, can't do this.

So, you might think that these two caveats would imply that more bits equal better resolution, at least in the DAC side of the process -- except that the error terms due to both of these issues are well beyond the audio spectrum!  In other words, you can't hear them.

A finite series of square waves can approximate ANY sine function (and thus sinc functions).  I believe it's been shown that as long as sine waves up to at least the 5th harmonic of the fundamental frequency are included in the series, any error terms again exceed the audible bandwidth.

There are similar results for the required amount of time over which summing occurs in the DAC.

Look, if ALL this were true, then CD would sound as good as SACD and DVD-A. 

CD does not sound as good as SACD or DVD -A.  Mastering comes into it, but in general, SACD and DVD-A does indeed sound better, so there is a hole in this argument somewhere.

Your argument that hearing cannot discern a denser bit stream is suspect.  There are engineers who do not agree with you.  A spectrum analyzer difference plot between CD and Hi Res analog output would show differences, and if those differences are between 20 to 20 Khz, then at some point, listeners will be able to hear them.

Newer DACs such as the SABRE 9018 can and DO sound different (and better) than older DACs.  The difference between a CD and SACD of the same recording can be heard.  Combine that with a discrete I/V stage, a good discrete output stage, and hi res sounds outstanding.