[sno1man]

Quote:

Originally Posted by Davey If it had gaps in the waveform then something was seriously wrong. You do realize that even low cost CD payers have extremely low distortion components, something that wouldn't be possible if there was any visible distortion like you're describing. 128K MP3 is highly compressed and far below the quality available from an uncompressed CD. I love vinyl, but don't understand your arguments.

I understand that and it was not well stated, but by its nature digital has less information in it than the analog original through sampling or whatever.

MP3 compresses and it samples. Both processes throw information away, of course the degree to which it does is highly variable.

We have a guy in our company that has an uncanny ability to identify a source as analog or digital even when the digital is at high sampling rates.

The other thing is that I think there are other types of distortion or sound coloring elements in both digital and analog, we just don't have the ability to measure them yet.

 

[Davey]

Quote:

Originally Posted by sno1man I understand that and it was not well stated, but by its nature digital has less information in it than the analog original through sampling or whatever.

OK, but that's a little different than claiming it has gaps. Even a LP has much less information than the analog original. So does a tape transfer. I doubt many would claim that a record sounds very close to the master tape, but run that master tape through a good A/D and most people wouldn't be able to hear the difference, except maybe your workmate

I'm not arguing that the CD sample rate or word length is good enough for high quality music storage either. It's not, although the quality of most pop recording and mastering today certainly doesn't exploit very much of what it is capable. But that aside, I was just commenting on the nature of your gap argument since I don't believe it's accurate.

 

[sno1man]

Davey,

Don't get me wrong. i think overall digital is a huge advance in audio overall. It just that it still has a ways to go and still does not completely capture everything that we had in analog (and I dont' mean, tape hiss or wow and flutter

)

And in analog your are right, there are considerable generation losses with every copy or transfer to another medium like LP to tape. That is one huge advantage of digital; a digital copy of a digital source can be perfect

But what I'm calling gaps is literally correct: when you sample something you break it in to discrete time based pieces. What a good dac does is essentially reassemble the pieces and fills in the gaps that got left out by making the sample, but it does that by assuming that the middle should be similar to what came immediately before and immediately after.
BTW: this exact discussion goes on just as much in the film versus digital realm

 

[Jigglybootch]

Quote:

Originally Posted by sno1man What a good dac does is essentially reassemble the pieces and fills in the gaps that got left out by making the sample, but it does that by assuming that the middle should be similar to what came immediately before and immediately after.

All DACs do that, not just good ones.

I think everyone's getting a bit off the subject, though. I was merely asking how could one call vinyl's ability to present high frequency spectral content an advantage over CDs when, in many cases, the speakers cut that content off anyway so that it never actually does get reproduced, and how is that any different from what CDs do, which is cut that high frequency content off at the source? I'm not sure that anyone has really answered that question. Personally, I don't see how there's any difference, and thus, for speaker/headphone systems that don't reproduce frequencies that will drive your dog crazy, you probably wouldn't notice that much of a difference (when it comes to hearing instruments/sounds that have such high frequency spectral content).

And my speakers don't have supertweeters, so let's not dwell on that.

 

[Davey]

Quote:

Originally Posted by sno1man But what I'm calling gaps is literally correct: when you sample something you break it in to discrete time based pieces. What a good dac does is essentially reassemble the pieces and fills in the gaps that got left out by making the sample, but it does that by assuming that the middle should be similar to what came immediately before and immediately after.

So I guess you no longer believe that there are gaps on the output of the CD player?

I don't know, Jiggly is probably right that this doesn't have anything to do with his post. But sample theorem isn't really intuitive. You don't really break up a signal into pieces with gaps between them. No data is thrown away as long as the signal and sample rate are within the prescribed constraints. Practical digital systems, especially CD, aren't ideal by any stretch, but the sample theorem has been proven over and over. Your comment about filling in the gap with assumed data seems to be directed at interpolating digital filters, but that's kind of another subject. They can be highly accurate, but I'll leave it there as they certainly aren't usually ideal filter implementations because of limitations of power and economy

 

[bigshot]

Anyone who thinks vinyl records contain more "information" than digital disks should watch a video on the old RCA CED "needlevision" format. That format attempted to store video on analogue vinyl and it sucked royally.

See ya
Steve

 

[s m @]

Quote:

Originally Posted by Davey Well, I love vinyl too, but characterizing digital sound as having gaps just seems wrong because I have many recordings that sound amazing, as I'm sure many of the people who post here do.

Sorry to flog a dead horse here, but I've been busy the past week. I never said I didn't like digital, I do like it a lot of the time, I have tons of CDs. But this is worth trying to explain again, b/c I don't think I did a good job earlier. But it's what I think is the main difference in character between digital and vinyl/analog:

People have used the word 'gaps', but we're not talking about tiny spaces of silence in the wave reconstructed by the DAC. What the DAC uses to reconstruct 1 second of music is 44K instantaneous samples of sound. Meaning that within that each 1/44K second, the sound changes not at all, and between each 1/44K section there is not silence but an abrupt change where one ends and the next begins. However 1/44K second is a VERY short time, so this still makes for pretty good sound. The best analogy I can think of to describe this, though, is how in an introductory calculus class, when they show you what an integral really represents (the area under the function). When the limit (sample rate) is low, you see a few rectangles approximating the smooth curve of the function, when the limit is 44K, you see 44K rectangles doing a very good job approximating the curve. When you take the limit to infinity the curve is perfectly rendered. Again, the sample rate of CDs now allows for good sound. However, to think that the way the sound is constructed does not affect its character is to say that a 90K sample rate would not be an improvment over a 44K one. Imagine how a recording done at 1 sample per second would sound! Analog has its 'infinite sample rate', but digital has advantages in other areas.

Quote:

Originally Posted by Jigglybootch I think everyone's getting a bit off the subject, though. I was merely asking how could one call vinyl's ability to present high frequency spectral content ...the speakers cut that off anyway...how is that any different from what CDs do, which is cut that high frequency content off at the source? I'm not sure that anyone has really answered that question.

I think people have answered that question, and that answer is 'You're right, but that's not really the reason most people hold up as the difference between formats', and then discussed what they/we think of as more significant differences than the freq range. IMO you're definitely right about that HF info being a dubious benefit (except to your dog).

 

[brian183]

Quote:

Originally Posted by bigshot The fundamental tone of cymbals don't get anywhere near 20kHz. And cutting lathes for vinyl had bandpass filters to prevent subsonic low frequencies from making the grooves too wide, or supersonic high frequencies from being reproduced as delicate modulation in the grooves that would wear down in the first couple of plays and turn into distortion. I believe that sampling rate is the argument in favor of vinyl, not frequency response. But I don't believe that makes anywhere near as much difference as mixing and mastering. The problem isn't the format... it's the engineering. See ya Steve

I realize im a bit late in this thread...but bigshot you took the words right out of my mouth. Although I've never REALLY got into vinyl, I've heard some CDs that are just amazing. The fact is, most people can't hear beyond the 20khz freq but I do believe vinyl has a certain...coloration/sound to it compared to a CD. You have to admit there are benefits of a format that is technically infinite compared to a format that is technically finite. Even so, the mastering is what really makes a good album/recording.
Just my two cents.

 

[Twombly]

Technically it's infinite resolution, but also more noisy.
Being able to hear 20 khz is actually really damned high.

 

[Davey]

Quote:

Originally Posted by s m @ People have used the word 'gaps', but we're not talking about tiny spaces of silence in the wave reconstructed by the DAC.

Well, when I questioned him about these "gaps" in the analog waveform, he said ...

Quote:

I have seen tests that show the gaps (like out of a CD player).

Quote:

Originally Posted by s m @ What the DAC uses to reconstruct 1 second of music is 44K instantaneous samples of sound. Meaning that within that each 1/44K second, the sound changes not at all, and between each 1/44K section there is not silence but an abrupt change where one ends and the next begins.

I'm not arguing that digital is better than vinyl or vice versa, or who can hear what. I've already said that I generally prefer vinyl playback, but I've also heard some amazing digital gear that (to me) exhibits none of the drawbacks you hear. Most DACs run at a sample rate of at least 352KHz and the DAC outputs current pulses that are then reconstructed into analog to whatever precision the designer deems necessary. The more filtering, the lower the distortion and the closer the signal approaches the original. There are no "steps" on the output of the CD player unless you listen directly to the output of the D/A convertor with no filtering (ie reconstruction). The output of the CDP is an analog waveform and the only way you can measure that it came from a bandwidth limited digital source at that point is by spectrum analysis to check the bandwidth. IMO, imperfect real time digital filter implementation due to economics and limited processor power is where the biggest problems lie with CD playback (other than the widespread poor mastering practices today as others have noted), not sampling theorem. although I certainly welcome the wider bandwidth available with higher sampling rates, especially since it also tends to make the digital filter problems less severe

 

[Publius]

Quote:

Originally Posted by s m @ People have used the word 'gaps', but we're not talking about tiny spaces of silence in the wave reconstructed by the DAC. What the DAC uses to reconstruct 1 second of music is 44K instantaneous samples of sound. Meaning that within that each 1/44K second, the sound changes not at all, and between each 1/44K section there is not silence but an abrupt change where one ends and the next begins. However 1/44K second is a VERY short time, so this still makes for pretty good sound. The best analogy I can think of to describe this, though, is how in an introductory calculus class, when they show you what an integral really represents (the area under the function). When the limit (sample rate) is low, you see a few rectangles approximating the smooth curve of the function, when the limit is 44K, you see 44K rectangles doing a very good job approximating the curve. When you take the limit to infinity the curve is perfectly rendered. Again, the sample rate of CDs now allows for good sound. However, to think that the way the sound is constructed does not affect its character is to say that a 90K sample rate would not be an improvment over a 44K one. Imagine how a recording done at 1 sample per second would sound! Analog has its 'infinite sample rate', but digital has advantages in other areas.

This digital explanation is wrong for modern DACs. Oversampling will reduce the period at which the DAC updates its output, so that an 8x OS will actually generate the waveform at over 300khz. The interpolation used for oversampling is very close to ideal (there is a bit of weasel room in terms of the finite FIR filtering used, but not much). After that you lowpass at 50khz or so and the upper 22khz frequencies are untouched.

Also there has been some speculation that the PVC in vinyl actually aligns itself in crystalline domains at the microscopic level, and that can effectively make the stored audio discrete-time at nearly audio frequencies. This effect would get reduced due to the pressure from the stylus averaging things out but it could still be there.

 

[sno1man]

Quote:

Originally Posted by Davey Well, when I questioned him about these "gaps" in the analog waveform, he said ... I'm not arguing that digital is better than vinyl or vice versa, or who can hear what. I've already said that I generally prefer vinyl playback, but I've also heard some amazing digital gear that (to me) exhibits none of the drawbacks you hear. Most DACs run at a sample rate of at least 352KHz and the DAC outputs current pulses that are then reconstructed into analog to whatever precision the designer deems necessary. The more filtering, the lower the distortion and the closer the signal approaches the original. There are no "steps" on the output of the CD player unless you listen directly to the output of the D/A convertor with no filtering (ie reconstruction). The output of the CDP is an analog waveform and the only way you can measure that it came from a bandwidth limited digital source at that point is by spectrum analysis to check the bandwidth. IMO, imperfect real time digital filter implementation due to economics and limited processor power is where the biggest problems lie with CD playback (other than the widespread poor mastering practices today as others have noted), not sampling theorem. although I certainly welcome the wider bandwidth available with higher sampling rates, especially since it also tends to make the digital filter problems less severe

Well I will admit to my knowledge possibly being out of date since the wave forms I was referring to were a few years back. When you looked at them on a high res scope they looked ragged for lack of a better term. I'm sure newer designs and things like oversampling help a great deal.

Whoever put up the integral formula to describe a curve got it right though. The more complex the math the more accurately it can describe the curve. That is in essence why I mean by gaps, missing information.

I do like digital very much though and believe it is a great step forward, but it still is not perfect in all cases. There is a certain "warmth" for lack a better term in some vinyl that seems to be missing in digital. Exhibit a for me is Bob Dylans Blood On The Tracks. Even the recent SACD which sounds better than the previous CD, just does not sound quite as good as the original LP.

Equipment plays in to this to. My analog set up is very high end, my digital stuff less so.

 

[Davey]

Quote:

Originally Posted by sno1man Well I will admit to my knowledge possibly being out of date since the wave forms I was referring to were a few years back. When you looked at them on a high res scope they looked ragged for lack of a better term. I'm sure newer designs and things like oversampling help a great deal.

That's fine, wasn't intending to pick on you or anything, just wanted to remind sm why I originally made that comment. With most CD players and DACs the output waveform is very clean, much cleaner than from any LP playback system, so I was just kind of surprised at your statement was all. Sure, there are some non-oversampling DACs that don't use any kind of filtering, or very little, and the waveform will have artifacts on the output, but that is far from the norm. Some people see those wav editor representations of the CD data words and seem to think that is what comes out of a CD player, but it isn't. That's just the data stored on the disc, displayed in a graphical connect-the-dots display format to resemble the actual waveform, but it doesn't represent reality.

 

[bigshot]

I remember back in hifi's heyday, there were "Test Records"... they would have tones on them that you used to see if your system was reproducing the full spectrum of sound. The frequency tones above 12k were EXTREMELY delicate. Usually, you would get a dozen or so plays out of them before you started to hear audible distortion from groove wear. High frequency modulation in the grooves is very delicate. The engineer doing the mastering would usually filter out super-sonic frequencies so the records wouldn't wear out too fast. Anyone who thinks that a vinyl record has significant information above 20kHz just plain doesn't know anything about records.

See ya
Steve

 

[jpr703]

Quote:

Originally Posted by s m @ People have used the word 'gaps', but we're not talking about tiny spaces of silence in the wave reconstructed by the DAC. What the DAC uses to reconstruct 1 second of music is 44K instantaneous samples of sound. Meaning that within that each 1/44K second, the sound changes not at all, and between each 1/44K section there is not silence but an abrupt change where one ends and the next begins. However 1/44K second is a VERY short time, so this still makes for pretty good sound. The best analogy I can think of to describe this, though, is how in an introductory calculus class, when they show you what an integral really represents (the area under the function). When the limit (sample rate) is low, you see a few rectangles approximating the smooth curve of the function, when the limit is 44K, you see 44K rectangles doing a very good job approximating the curve. When you take the limit to infinity the curve is perfectly rendered. Again, the sample rate of CDs now allows for good sound. However, to think that the way the sound is constructed does not affect its character is to say that a 90K sample rate would not be an improvment over a 44K one. Imagine how a recording done at 1 sample per second would sound! Analog has its 'infinite sample rate', but digital has advantages in other areas.

I would agree that fluidity, not tonal range is the driving force here and the quoted post proivdes a pretty good explanation. After all, the pursuit of a more fluid sound wave has been the driving force behind lots of new hi-fi pursuits like high rez formats and upsampling. To me the fluid sound of vinyl is the driving force behind what people refer to as PRaT. Those smoother waves just sound more realistic and have a way of drawing you into the music and getting your foot tapping. Maybe it's a personal thing, but listening to music with lower sampling rates just doesn't do that for me.

None of this is to say that vinyl is perfect. Pops and hiss can be even more distracting than low sample rates, but I think that people are beginning to realize that digital does need some smoothing out to sound truly authentic.