Talk:Myths (Vinyl)

From Hydrogenaudio Knowledgebase
Jump to: navigation, search

"The term "analog", by definition, means that the signal is not and cannot be a perfect reproduction of the original - it is merely an "analogue" of the existing signal, corrupted in the process of encoding."

This is not accurate, analogue has two definitions. One is a noun and refers to "a person or thing seen as comparable to another", the other is the adjective form that has nothing to do with the noun. [wrong - see below-Tnargs] In the world of science it is the adjective form that is used, which means "relating to or using information represented by a continuously variable physical quantity rather than digitally" [Wrong - the OED says "relating to or using signals or information represented by a continuously variable physical quantity such as spatial position, voltage, etc." Also note that the noun and adjective are closely related - Tnargs]

These definitions are from the Oxford English Dictionary, retrieved from here: [Wrong - it is the Collins Concise. I have edited the above with the OED definitions - Tnargs] In fact I submit that this entire section: Vinyl is better than digital because the analog signal on the vinyl tracks the analog signal exactly, while digital is quantized into steps of the article be deleted.

The logical foundation of the argument is faulty at best. Vinyls are not better than CD's as far as objective terms can dictate, but analog is fundamentally and demonstrably more accurate than digital. I will now demonstrate: [The 'demonstration' below is wrong in almost every way, just like the above claim regarding accuracy. - Tnargs]

If one were to record a sound wave simultaneously into an ideal analogue audio signal and an ideal digital audio signal, then the analogue signal will always be a more accurate reproduction than the digital one. [Your use of 'ideal' is naive and unhelpful. An ideal analogue recording has infinitely low noise and distortion, and an ideal digital system has infinite bit depth and infinite sample rate. Discussing such an example obviously gets us nowhere: it is not useful to discuss it. Removing the word 'ideal' from your claim, it is only true at digital bit depths of less than about 20 -- at higher bit depths the digital overtakes the analogue for accuracy - Tnargs]

No matter how small the quantization of the digital audio would be, it would still be separated into discrete variation, where the analogue would, again, be unseparated, as a continuous variation wave. An ideal analogue recording could theoretically produce an exact electrical representation of the mechanical energy of the sound wave, where even an ideal digital could not. [This is only true if the analogue signal has infinitely low noise, which is just as impossible in practice as a digital representation with infinite bit depth. Both are equally silly. If you consider any real-world analogue noise level, then there is a number of bits (around 20) where the digital accuracy exceeds the analogue accuracy. - Tnargs]

Furthermore analogue recording does not "encode." Digital audio signals replicate audio through a series of "code," that is, they represent a mechanical wave of air pressure in a "code" of information, that is then read by some decoding instrument that can produce sound waves. [There is nothing true here. Think of digital audio as a temporary storage/transfer medium, between analogue in and analogue out. It is important to remember that the output of a digital playback system is an analogue waveform, not digital. The only thing you need to compare are the three analog waveforms: [1] original analogue input; [2] analogue output through an analogue storage/transfer medium; [3] analogue output through a digital storage/transfer medium. With modern PCM audio, you will always find waveform 3 is closer to waveform 1, with waveform 2 coming last in terms of accuracy. - Tnargs]

Analogue on the other hand uses a transducer to convert the mechanical energy of the wave into electrical energy, and possibly into magnetic energy in the form of "tape," and then uses another transducer on the output end of the system to convert back from electrical or magnetic energy into mechanical energy, in other words, a sound wave. [And the magnetised tape inevitably behaves as an audio compressor and limiter with non-linear curves that do not occur in digital audio (although one can replicate tape curves in digital if one wishes to be less accurate than with straight digital). - Tnargs]

In fact, returning to the original quote, Digital signals are actually disqualified from exact accuracy by their definition. To be digital by definition means to be sampled, or quantized, or represented as discrete, rather than continuous, variation. [See above explanations. Your mistake is to look at the digital data as a 'waveform' instead of examining the analogue waveform that is created from the digital data. - Tnargs]

I do not disagree with the overall point of this article, but if you take vinyl as nothing more than analogous to analogue audio signals in general, then it has an accuracy advantage over any digital signal, regardless of any means taken to mask the quantization of wave variation, including dithering or any kind of sample blending. [Complete nonsense. See all my comments above. Vinyl is very, very much more than 'analogous to analogue audio signals in general', and I am referring to 'more' in terms of errors. The multitudinous and gross errors of vinyl are too numerous to go over here, but they comprise not only gross noise and distortion additions, but actual signal deformations too. Modern PCM audio has ENORMOUS accuracy advantages over vinyl. - Tnargs]

The section seems to be reaching for advantages for digital formats by intentionally confusing the issue, in an article that already proves basically without question that present digital audio signals are objectively superior to present analogue signals. [It is you who are confused, I am afraid. - Tnargs]

I will check back in a day to see if a counterargument has been posted, and then will remove the original sentence for factual inaccuracy.

As an afterthought, some of the definition confusion may be coming from the fact that we spell analogue without the last two letters in the United States when we refer to the adjective form, but still include the "ue" when using the noun form.

[The above section is almost riddled with gross errors of fact. I have only begun to correct it with my comments above. It certainly should not be considered as a reason to edit the main article. -Tnargs]


After seeing the retractions and additions that were made, I recognized that an expert contribution was required, since one inaccuracy was replaced with another. I decided to revert and start from that version as base, but essentially duplicated the retractions anyway by the time I was done.

In recognition of the above objections, I decided to update the section to reflect modern digital usage patterns that more sharply delineate the technical advantages of digital storage and processing versus analog. Regarding the prior request for technical corrections to the quantization verbiage, I left it as is because it is a great layperson-friendly explanation, and I could see no way to improve it without damaging it, so I just elaborated a little around the edges to explain better (I hope).

ps Regarding the assertion on this talk page that 'ideal analog' and 'ideal digital' recordings differ in fidelity, one must keep in mind the following two facts:

1) An ideal digital recording has infinite sample rate (bandwidth) and infinite bit depth (dynamic range) and infinite sampling accuracy (zero noise or distortion), approaching in the limit an exact tracing of the original signal, just as an ideal analog recording likewise has infinite bandwidth and infinite dynamic range and zero noise or distortion, approaching in the limit an exact tracing of the original signal.

2) There is no such thing as an ideal recording, not even an ideal analog recording. Technical reasoning must always proceed from real data and real theorems in order to end up with real results.

Just a slight correction... digital accuracy exceeds analog accuracy well below 20 bits. If it is accepted that bit depth defines the dynamic range and the related signal to noise ratio, then working back from these measurements can demonstrate that studio quality 2" analog tape running at 30 ips has equivalent measurements to about 13 -14 bits, without dither. Vinyl can vary from 10 to 13 bits depending on the frequency and the position on the LP, while analog cassette tape is about 8 bits. See for references.

In "Vinyl sounds better than CD", I think you have the terms subjective and objective reversed. Objective means perception so it take precedence. When people look at numbers they can ignore objective perception and thus becomes a strong bias that does not match reality. Unless a real explanation is made for the discrepancy between perception and numbers then I have to assume the numbers to be wrong, missing, or misinterpreted. [Incorrect, subjective means relating to feelings/taste, objective is without feelings/taste. For example 10km race is longer distance than 100m race this is objective; where as 10km race is harder than a 100m race is subjective. Note that more people are likely to feel differently about the second statement than the first. Same with music, you may feel that one format sounds better than another, and that is your subjective opinion and is your own personal 'truth', but is not universally true. You might also beleive that a 12" vinyl is smaller than a 12cm CD, however you would just be wrong with no wiggle room as this is an objective measurement. Now here the objective measurements are saying in these aspects that the digital is closer to the original than the vinyl; there is then a jump in some people's minds that mean that digital is better than vinyl, There maybe other objective measurements not taken/understood that have a greater weighting on perception that aren't yet known about. The weighting of such objective observations would be subjective.]

Something I find odd when people talk about Nyquist frequency, this is the sample rate where a frequency 'can' be represented, however this ignores phase of the recorded wave and assumes in phase. So if your amplitude is 1 for 20kHz and in phase cosine wave with a sample rate of 40kHz, then given perfect ADC and DAC processing, it has amplitude 1,(I.e. it stores 1,-1,1,-1,...) now knock the phase by a quarter cycle and the amplitude drops to zero (I.e. stores 0,0,0,0,...). Now take a sample rate of 160kHz and the worse case scenario is 10xlog(cos(22.5deg)=~ -0.34dB. Which to me seems like a nice practical limit where subjectively I doubt people could benefit from a higher sample rate. However at sample rate 44.1kHz where these frequencies with a slowly changing phase (likely in practical scenario with real instruments and mics), is it not possible for vinyl to be more faithful starting somewhere north of 5.5kHz upwards to top of normal hearing, in a way that could be subjectively heard as better?

Has there been any proper studies comparing difference between vinyl full analog path from cartridge to speaker coil, compared to putting high end realtime ADC and DAC into the same chain? My personal hunch is vinyl's tend to be mastered 'better' for audiophiles and vinyl's tend to have less high frequency transients, that can be fatiguing.