Talk:Comparison of analog and digital recording
{{WikiProject banner shell}}
to this page and add the quality rating to that template instead of this project banner. See WP:PIQA for details.![]() | Professional sound production Start‑class Mid‑importance | |||||||||
|
![]() | This article was nominated for deletion on 2006 May 30. The result of the discussion was keep. |
|
Was it ever entirely analog or digital?
This section makes the claim that the human ear is an analog-digital hybrid system (without claim, but that is aside from this point), and then two sentences later says that considering any part of the ear as acting digitally is a misconception.
I would say that this section could benefit from editing, but really, it is just a conversational piece, with nothing to back the claims. While it is interesting information, it might just be best to delete this section, lest someone wants to completely rewrite it. 72.29.164.238 (talk) 00:16, 2 May 2014 (UTC)
Suggest inclusion and (minimal) discussion of 2007 Moran/Meyer AESJ paper
Copy available at http://drewdaniels.com/audible.pdf. It debunks Stuart (cited in this page) and similar claims via blind testing of hi-rez passed through RBCD bottleneck; no one was able to detect this alleged degradation. — Preceding unsigned comment added by 72.93.248.186 (talk) 00:11, 29 December 2013 (UTC)
Suggested extension to the section on Analog Warmth
Philosophers who have addressed the issue of "analog warmth" point out that its source may result from unique imperfections that appear on only one copy of an analog LP (a token of a type), played on a specific system, and at a particular time wherein the entire environment works to create an irreproducible moment. This moment may not transmit the live performance "accurately" and is considered by the musicologist and philosopher Theodore Adorno a distinct esthetic experience (cf Current of Music published by Polity international and Stanford University Press, but I'll be damned if I do you convenience store clerks your work for you any more than I goddamn well have to). A digital component would reduce the uniqueness (what Adorno's friend Benjamin called the "aura") mathematically.
In riposte to claims that such nuance is "beyond the range of human hearing", Adorno's riposte, or at least the printable part, would be that "studies" of "human" "hearing" use empirical subjects either selected or self-selected out of a normalized population which would, under actual conditions of production, not include audiophiles and people with superior hearing.
I seriously think the above should be included because the rest of the article is mostly bullshit scientism, typical of wikipedia's borderline retarded output.
Edward G. Nilges Hong Kong 4 Jan 2012 — Preceding unsigned comment added by 203.218.45.1 (talk) 16:36, 4 January 2012 (UTC)
- Stalking Edward from the sopa article, he seems to have correctly identified some problems, and my god this article sucks on approachable language, it's on my mental list of things to do. Anyhow Edward, I'm not so sure such flowery language is better than the audiophile and engineering jargon that's crammed the article full, but if I can find it, I'll include it. Maybe a quick comparison to 'old fashion lightbulbs' (incandescent ones), pointing out that whilst they don't render colors so perfectly, they indeed impart warmth especially to people and portraits, restaurants, and so forth, because like analog recording, or at least LP's, their dynamic range is so characteristic. Penyulap talk 19:51, 24 January 2012 (UTC)
I am sagar?
I'm sure I'm forgetting the right place to post this question, so I'll stick it here. Is there any clear "winner" in this discussion, or, as I suspect, is it all simply preference or (possibly imagined) advantages? If the former, I couldn't find it in my twenty minutes' reading of the article or in an hour's perusal of Audiophiles. If the latter, could someone create a clear list (succinct as possible, I know it's difficult)of pro/con's for each? Very interested in the topic, any help on issue would be greatly appreciated. Thanks "Are you suggesting coconuts migrate?" (talk) 06:40, 25 December 2009 (UTC) 32Lex22
Wikipedia's articles are meant to be neutral. Did I solved ur confoosion ^~^? Ivaneduardo747 (talk) 01:39, 23 March 2011 (UTC)
- If you look at sales of recordings and recording equipment, digital is the clear winner. If you want to know which sounds better, it depends on who you're asking and what they're listening to. --Kvng (talk) 16:57, 24 March 2011 (UTC)
Misstating what "analog" means
While the explanation given (direct correspondance to a physical property) may have originally have been true (I will not presume to judge this), it is not in accordance with modern usage. The difference between analog and digital can, with some oversimplification, be stated as the difference between a continous dependency and one that considers (e.g.) a signal a series of discrete information units. (This statement is off the top of my head and may be sub-optimally formulated; the principle is still correct.) A physical dependency is only needed in as far as everything (including the digital signals in a computer) ultimately has some physical connection. 88.77.150.159 (talk) 02:09, 22 February 2010 (UTC)
There is some confusion regarding Nyquist sampling. This is applicable to cw sine waves that are time invariant, i.e. continuous with time. In this respect it is accurate as stated by the theory. However, this would make for very boring music. Luckily, most music involves dynamic transients that result in a time envelope around the fundamental frequencies of the instruments. The fundamental note may only be at a modest frequency of a few kHz. However, to reproduce the time attack and decay of the note accurately can require much higher sampling frequencies to reproduce the time variation of the waveforms. To reproduce these requires sampling the waveform at a time resolution of atleast half the finest transient time that is needed to be reproduced. The other information that is needed to provide an accurate stereo 3-D soundstage of the music is the correct phase of the frequencies. i.e. It is not enough to reproduce a 10 kHz signal, but its time alignment with respect to the other frequencies in the overall signal needs to be correct. Simple Nyquist sampling does not have enough information to reproduce this. This provides the spatial placement of the instruments that adds to the interest and accuracy of the music reproduction.
The specifications for typical casette decks are too pessimistic in the article. Typical midpriced casette deck available in the 1990's such as a TEAC V-1030 provide a usable SNR of about 70 dB with Dolby B and 80 dB with Dolby C. The frequency response of three head casette decks can extend from 15 to beyound 21,500 Hz at -10dB to -20 dB levels and to better than 20,000 Hz at 0 dB with suitable tape. Please note that this is +/-3 dB and thae actual usaable frequency response extends further into the frequency extremes. By adjusting the tape bias for a given tape formulation, very good reproduction of vinyl LPs or Bluray soundtracks can be provided with extended high-frequency response. —Preceding unsigned comment added by 207.35.47.66 (talk) 22:57, 3 May 2011 (UTC)
- Your statement regarding Nyquist sampling is mostly correct. It applies to periodic, time invariant signals, which does NOT include audio recordings. For the time invariant signal you must sample at twice the highest frequency component in order to reproduce exactly. For signals that vary in time you must sample considerably higher, like 4-10 x the Nyquist rate. That the 44.1kHz sampling rate of a typical audio channel works as well as it does is only because the frequency components of audio above about 6kHz are quite small most of the time. Music having a synthesizer generated signal can contain significant signal above 6kHz, though. DSP's that handle comm signals need to sample much higher than the Nyquist frequency in order to properly recover the signal. You are also correct in stating that this (44.1kHz) undersampling produces phase and amplitude distortion in the reconstructed signal that affects the perceived spacial placement of the instruments, especially at high frequencies.71.214.218.217 (talk) 19:15, 9 August 2012 (UTC)
- I have reverted uncited changes you've made to the article. Your assertions are not supported by Nyquist–Shannon sampling theorem or Sampling (signal processing). --Kvng (talk) 00:21, 12 August 2012 (UTC)
Recent experiences
I came unto tube amps quite accidentally five years ago; I found an early-60s Sherwood tube stereo (possibly the last one before the change to transistors). I dug out a SONY turntable, two Yamaha natural speakers (also found), pulled an MJQ record. (Even before the amp warmed, my cat became very interested in the sound coming from the needle; to this day, he only "smiles" for the MJQ.) When it warmed, it was as if the long gone quartet was in the room -- it was not like listening a good stereo (such as my AV 85 PY), but as if they were actually in the room. Over time, the driver tubes gave out and were replaced with guitar amp tubes, and then other problems including a ceramic resistor (mechanical) breakdown put the Sherwood into storage.
I think that the 60-70s hunger for music drove the conversion to transistors, as everybody needed music (or other sounds) to learn from to perform and not necessarily appreciate such as audiophiles of the 50s-60s had. I know that I was satisfied with second-rate recordings of African and Caribbean music from distant radio stations recorded on salvaged tapes with salvaged recorders, as I wanted the sounds to learn from.
Getting to the heart of the matter -- I attempted to record these fairly distorted but usable tapes to my laptop, and the sound going from the tape deck through the computer (audacity) and to the solid state amp was so much further distorted that it because useless. When piped directly to the amp, it is actually pretty good, which is what I am listening to as I write this (perhaps the only recordings of college station gigs to survive).
I am not seeing any material actually describing what I am experiencing; and, no doubt, there may be a rush of what I call "AudioFOS" to attempt to "protect" the existing material here promoting digitalization.
I attempt to abstract tube amplification (as ArtSci) so that I may someday engineer tube amps, as I want the MJQ back in my living room.
Another example of how digital/solid state weakens music is Jimmy Guffre's third-stream Serenade, which I believe is the finest piece of performed music (through the Sherwood tube amp from a record recorded with tubes), sounds labored and weak in the Amazon MP3 version.
It may be possible that I would benefit from a high-dollar digitizing device, though I cannot (yet) find material suggesting this (most material supports sound cards), but I really don't see spending money on such a thing when I will benefit most by sticking with what I have (including an also-found early-90s Technics cassette player).
Also, I believe "warmth" is a red-herring (or more audioFOS). --John Bessa (talk) 17:24, 17 February 2012 (UTC)
Katz research and conclusions about low-pass filtering
Added a paragraph relaying his results in his book that suggest variance in sample rates is due to pre-A/D conversion filter design. Feel free to discuss. Radiodef (talk) 18:16, 13 September 2012 (UTC)
- Nice addition. I'm not convinced this directly relevant to the A vs. D topic of the article; It's more of a D vs. D discussion. The whole section might want to be merged into Sampling_(signal_processing)#Sampling_rate. --Kvng (talk) 21:30, 13 September 2012 (UTC)
- The section on higher sample rates in this article could be better used, I think. Some say that analog is better than digital because its frequency response is wider and "more detailed". In the case of analog tape, for instance, 15IPS goes up to 30kHz, and exhibits a fall-off slope, like a low-pass filter. At higher frequencies with digital, the wavelengths are short enough that there isn't much space between the positive and negative ends of a cycle. Figures e-g of the second image here show what I mean, though the chapter in this book is about something different:
- http://www.dspguide.com/ch8/4.htm
- There are various reasons that this could mean that analog performs better, and I think that's what the higher sample rates section is trying to get at. Technically, from a mathematical perspective, no detail is really lost in the high frequencies of sampled audio, but it looks that way. Radiodef (talk) 16:12, 14 September 2012 (UTC)
- The narrowness of the record and playback head gaps of the tape recorder combines with tape speed to set the high frequency response limit on tape. Bias signal has to be jacked up for narrower head gaps, but too much bias reduces the high frequency headroom. (It's not just tape speed.)
- There's a practical human limit to all of the comparisons: most people cannot hear much above 15kHz, and even a Golden Ears person has a high frequency response that is quite a bit less sensitive than the range between 1k and 4k. Binksternet (talk) 16:34, 14 September 2012 (UTC)
Reducing the information in this article
There are a lot of really long paragraphs in this article. The information is all good, but perhaps too comprehensive for an overview like this. I might go through and try to shorten some sections. To me, it seems like the people who will be reading this article are looking more for a presentation of the performance differences, and less for technical details about how they work. There are already individual articles that have comprehensive details about the formats, and this article is supposed to be a comparison of the two. There's also a note that the neutrality is disputed in this article. Not really sure why. The article just presents the facts and doesn't really argue one way or the other about which is "better". Radiodef (talk) 18:43, 13 September 2012 (UTC)
- I've made some changes. The article really needs more sources too, and is pretty filled with weasel words ("some", "many", etc). Wording could use a lot of improvements. Radiodef (talk) 19:32, 13 September 2012 (UTC)
- Good work so far. Any reduction in useless wordiness is, in my view, a practical increase in information because the article is more easily read.
- We could use better sources, too. John Eargle's Handbook of Recording Engineering and Audio Engineering for Live Sound are excellent books. One relevant section is on page 44 of the latter book: "Is digital always better than analog?" He says an outstanding advantage of digital audio is in signal processing, an advantage that is not always utilized by the consumer but is ever present during audio production. The article says almost nothing about that. Bruce and Marty Fries have a good book in Digital Audio Essentials. Binksternet (talk) 20:09, 13 September 2012 (UTC)
- I hinted at it at the top, where I added a comment about "more transparent filtering", and I'll probably clarify that. You're right that the article doesn't really say much about DSP advantages, one obvious one being linear-phase equalization. That's a great source, too. I may just add a section on this. I added a bit about analog modeling plug-ins, and it's a little out of place where it is now, and could go in that section too. Radiodef (talk) 16:22, 14 September 2012 (UTC)
- I agree. Shorter and verifiable is what we want. Some of the recent edits have added new uncited information and that's not the right direction. --Kvng (talk) 21:23, 13 September 2012 (UTC)
- A lot of the larger paragraphs in the beginning would be a lot better as a main article/summary. It also looks like the article was originally written as arguments for analog --> counter-arguments, but it's been convoluted along the way and is a little confusing now. I don't really think that is the right tone anyway since the article is called a comparison.
- I think the article could be fairly long since there's a lot of information to cover, as long as it's concise, and right now the layout is vague. Radiodef (talk) 16:22, 14 September 2012 (UTC)
- Lots of Wikipedia early-written articles end up looking like a patchwork quilt; this one is a prime example. The first version of this article, though it was terribly insufficient and though it made a wrong assumption (mechanical impression on magnetic tape?) it was surprisingly ambivalent about which format was better.
- Such patchwork articles are superb candidates for a complete rewrite by someone with a lot of good references and a solid focus. Throw out the wobbly parts and keep the good wood. Binksternet (talk) 16:43, 14 September 2012 (UTC)
- Yep. And I think this article is also relatively important since the subject is so often debated, sometimes viciously. I can revise the information about digital audio pretty heavily, but I don't know enough about analog technology to rewrite those parts confidently. Radiodef (talk) 17:10, 14 September 2012 (UTC)
- There is a subsection titled "Digital fundamentals" under the "Noise and distortion" section. This subsection doesn't really have to do with noise, so much as it is an overview of how digital audio is sampled and stored. Could someone who knows more about analog technology write a similar (but shorter) subsection that summarizes how analog technology works? Then both can be put under a new section, each with respective Main Article redirects, and I will shorten "Digital fundamentals". Radiodef (talk) 17:04, 14 September 2012 (UTC)
- The quicker solution is to throw out the "Digital fundamentals" section. The only relevant information would be a direct comparison of this or that feature of analog vs digital recording. There is no need for a whole section on each format. Binksternet (talk) 17:36, 14 September 2012 (UTC)
- We could do that too, but a brief overview of sampling technology is needed for elsewhere in the article, mainly quantization. We could toss it and put a very brief overview (3-5 sentences) in the quantization section. That'd work for me. Radiodef (talk) 18:34, 14 September 2012 (UTC)
- Done, and done. Radiodef (talk) 19:51, 14 September 2012 (UTC)
New section: Signal processing
I've started a new section, titled "Signal processing". I put some information in there about analog gear and digital filters, and moved my paragraph about analog modeling there as well. I don't know much about vintage analog processors, so that could easily be expanded. This section at Digital filters could also be linked somewhere, though it's a technical comparison of filter design, not recording equipment. Radiodef (talk) 19:57, 14 September 2012 (UTC)
New Section: Missing Link in the Recording Chain
In audio recording, digital or analog, instantaneous air pressure is converted by a microphone into a varying voltage, or an analog electrical signal. In the "analog" recording process, that voltage is used (in cutting a record) to move a cutting head, whose position is only an approximation of the voltage, subject to constraints of mass, velocity, resistance, and more, not to mention the effects of local magnetic fields, vibrations, temperature, air pressure, humidity, and more. In recording to tape, there is a similarly long list of potentially distorting effects.
In the "digital" process, the measured voltage in each instant is represented with a number whose accuracy is limited by the number of bits used to represent the number, the accuracy of the voltage-to-number (analog to digital, A/D) converter, and the rate at which samples are taken. The A/D converter's accuracy may also depend on a number of factors, including manufacturing processes, materials variation, temperature and more.
Modern recording processes utilize a staggering level of precision in these processes, analog or digital.
Hence, arguments that the analog process somehow captured some characteristic of the sound that the digital process missed are at best fanciful, at worst exploitive.
That is not to say that analog recordings sound exactly the same as digital recordings, as all the effects mentioned above surely alter the recorded waveform (and likewise for playback) in a way that some people might indeed prefer. Just as the sound of an electric guitar is often preferred with a measure of distortion, so too might be a musical performance be preferred with specific kinds of distortion. But if this is the case, let us be clear that it is the distortion, not the greater accuracy of reproduction, that is responsible for the preferred sound. Nancy N (talk) 03:06, 16 November 2013 (UTC)
- Are you proposing to add this to the article? ~KvnG 00:14, 20 November 2013 (UTC)
this article really sucks as s. thank you.
a — Preceding unsigned comment added by 49.187.168.119 (talk) 21:31, 2 May 2014 (UTC)
Frequency response of master recording
I think the essential thing that is missing is the drequency response of the first step of the recording, the multitrack master reels and the ginal mised tape master. Anyonw can add anything?