Talk:Digital audio

From Wikipedia, the free encyclopedia

This article is within the scope of the Professional sound production WikiProject, a collaborative effort to improve Wikipedia's coverage of the technology, equipment, companies and professions related to professional sound production. If you would like to participate, you can visit the project page, where you can join the project and see a list of open tasks.
B This article has been rated as B-Class on the assessment scale.
Society Portal
This article is part of WikiProject Media, an attempt to better organize information in articles related to media. If you would like to participate, you can edit the article attached to this page, or visit the project page, where you can join the project and/or contribute to the discussion.
The references in this article would be clearer with a different or consistent style of citation, footnoting, or external linking.

Contents

[edit] Definition of Digital Audio

I think this needs a bit more clarification. It seems many people are still confused, and believe that it is possible to listen to a digital audio signal. It is not. Your eardrum cannot process digital information. It can only hear an analog sound waveform. Digital audio is merely a representation of this analog waveform. I think the article should be more clear in the fact that digital audio is merely a carrier between the original analog sound source (maybe a voice speaking into a microphone) and the analog sound coming out of your speaker hitting your eardrums. 137.71.23.54 17:05, 8 February 2007 (UTC)Brett

[edit] Advantages of Digital Audio

Perhaps this should be expanded to include some other things such as: Increased processing capability (some effects, including various kinds of delays, filters, modulation, and psycho-acoustic enhancement, are either very expensive, difficult, or impossible to implement with analog components... but can easily be implemented in DSP); Non-destructive editing and processing; Immunity to noise and distortion while in the digital state (although we all know D->A and A->D conversion has these problems); Project sharing among different studios; Error Correction in case of media corruption, etc. These are all very objective and certainly non-biased advantages of digital audio. 137.71.23.54 17:05, 8 February 2007 (UTC)Brett

The biggest advantage of any digitial system is the ability to store, retrieve and transmit signals without any loss of quality. The current wording doesn't really emphasise this. Also important is the fact that it was driven by a need to introduce digital audio into broadcasting systems in order to minimise degredation from long and complex signal chains.

Some of the advantages that you mention are true of course, but I think they would be better categorised as "applications" as they cloud the main factor which leads to all of those things. For example, DSP was not a motivating factor and proponents of analogue will disagree that digital processing really offers anything of value - the majority of effects used will have analogue counterparts. Error correction is not an advantage, it's an absolute necessity e.g. Reed–Solomon on CD.

I know it's mentioned later but maybe it should be noted that by its nature, digital audio inherently represents a less than perfect view of the original analogue signal.

Basically, I think this section should also answer the question "why not stick with analogue?".--Subtractive 22:32, 13 February 2007 (UTC)

That depends on what assumptions you place on the analog signal. No analog system can carry an infinite bandwidth signal and at some point the signal will hit the noise floor and you've effectively found your upper frequency. :) "Perfect" is a misnomer anyway.
I think with regards to DSP having an advantage is that it's software. Being able to reflash a new algorithm to give you a completely different function has an incredible value/advantage over doing it in hardware. I don't think this point is really that contentious either. How many of people do you know use punch cards to interace with a computer? Cburnett 22:48, 13 February 2007 (UTC)

[edit] Professional audio industry

The professional audio industry is moving to 96 and 192k sampling rates. Another important issue is bit depth, many consumer devices are only 16 bit while professional application are moving to 24 bit depth.

→In response to this I'd say the process has already been completed. Studios running under 192kHz and 24-bit are almost gone at this point. Even though the end media is usually 44.1kHz, 16-bit (standard audio CD), recording in 192-24 dramatically reduces the noise propagated through the system and increases the reproducible frequency range, which can yield a higher quality master, even after downsampling and truncation. 137.71.23.54 17:05, 8 February 2007 (UTC)Brett

[edit] Diagram request

Would it be possible to get an image of the digitization process for audio - shows the waveform and how it's reconstructed digitally... I've seen them before but I haven't been able to find a diagram that's under a "free" license.

This [Handbook Sample chapter] has diagrams like what I'm envisaging, but it's definately copyrighted! VoltageX 23:47, 7 August 2006 (UTC)

→I'll make one if someone will host it. 137.71.23.54 17:05, 8 February 2007 (UTC)Brett


Brett, I wish I'd seen this earlier - you can upload directly to Wikipedia, no hosting needed! VoltageX 02:46, 2 June 2007 (UTC)

[edit] Accuracy of the Digitization Process

Although the math is covered in the Nyquist–Shannon sampling theorem article, I think it's worth a mention that in a theoretical discrete-time system (not digital, mind you, just discrete-time), a signal can be perfectly sampled and reconstructed as long as it is band-limited by the Nyquist rate. Now, in real life there are higher harmonics we'd like to capture, and when the amplitude is quantized we get quantization noise, but I still think the Nyquist theorem should be highlighted here. 137.71.23.54 17:05, 8 February 2007 (UTC)Brett

[edit] MIDI and digital audio

In the "Digital audio interfaces" section of this article, it lists MIDI, and states it as "low-bandwidth interconnect for carrying instrument data; cannot carry sound". Now, if I'm not mistaken, this is not exactly true, since MIDI can carry digital sound via SYSEX dumps, no? Such wouldn't really be totally legit though, I would assume, due to the SYSEX capabilities of MIDI varying from manufacturer to manufacturer to carry any kind of binary data (hence it's name, SYStem EXclusive)... misternuvistor 14:09, 1 October 2006 (UTC)

[edit] Sysex carrying Audio

Whereas it is true that audio can be sent via MIDI Sysex, it is certainly not able to do this in real time. A single CD quality monophonic audio stream will run 16 bits at 44,100 samples per second. That translates to 88,200 bytes per second or 705,600 bits per second. Sysex sends data in packets, maximum bit depth of 7 bits. Since MIDI is running at 33,100 bits per second, and each packet consists of a start bit, a flag bit, seven data bits, and a stop bit, the maximum audio transmission will be .7 * 33,100 bps, or 23,170 bps. Thus it will take a minimum of 30.5 (rounded) seconds to transmit a single second of monophonic CD quality sound. Stereo doubles this, and therefore we get the working value of about 1 minute of data transmission per second of audio.

Obviously one can transmit audio data in many different formats: one could stack pennies in head=0 tail=1, then roll them up and carry them across town. However, I would not recommend this as a mode of transmission! Davidbrucesmith 02:54, 24 October 2006 (UTC)

[edit] Subjective evaluation

I think that this section should be revised. Placing proper controls in subjective tests makes the results repeatable which is essential in any scientific measurement. It should be made clear that subjective values given like detail, body, etc. can be measured in controlled subjective tests. This can take the form of a rating given across a numerical scale. For example detail or tonal quality might have extremes of too bright or too dull. These evaluations are useful because of their repeatability and the fact that the results can be analysed quantitatively, using, for instance, statistical methods. —The preceding unsigned comment was added by 217.205.242.230 (talk • contribs) 14:59, 3 December 2006 (UTC).

Indeed. I wrote the bulk of this article for similar reasons as why this section needed revising. Cburnett 20:49, 3 December 2006 (UTC)

I'm thinking of adding some further information on the common subjective tests. These are the double-blind tests like ABX and open tests which are often used in the audio press. I'll explain the advantages and disadvantages of both methods. Floyd Toole of Harman International has written on this subject.

In The Loudspeaker and Headphone Handbook, Floyd Toole states that objective measurements of loudspeaker performance, particularly amplitude response, are a good indicator of probable evaluation of subjective performance. I think that this link, in broader terms, is also true of other parts of the reproduction chain.

The point made in the second paragraph of the 'Subjective evaluation' section -

Audio can be measured and analyzed to more exacting measures than can be done by ear, but what this technical measurement and analysis lacks is the ability to determine if it sounds "good" or "bad" to any given listener

is true, but only in the sense that (objective) measurements won't necessarily correspond to the listener's enjoyment (subjective rating) of a particular performance. This is different because whether or not a listener enjoys a particular performance or not depends on many things, like the production of the recording, and the listener's state of mind. One of the advantages of double-blind testing is that psychological controls attempt to reduce the influence of unrelated factors - unrelated in the sense that we are only interested in the performance of the reproduction equipment, and not the emotional/psychological state of the listener which is quite variable independently. There is also the adaptation of the senses to consider. This alters a person's perceptions of a stimulus over time. To counteract the effect of adaptation, double-blind tests are usually conducted over short time periods. A disadvantage of double-blind testing is the difficulty in getting high-quality data which are able to differentiate high-performance equipment. --Enescot 12:59, 16 December 2006 (UTC)

I would recommend reading over Analog sound vs. digital sound and going from there. Cburnett 22:35, 16 December 2006 (UTC)

→I'm having trouble with this subjective portion. What you think sounds "good" generally depends on what you've listened to in the past. For example, if you grew up listening to classical music you're going to expect a huge dynamic range, while people growing up listening to punk will expect a very compressed sound. It's just what you're used to. Things such as recording media, amplifiers, and speakers, can all affect these sorts of parameters. So systems that have some weird compression might sound "good" to a hard rocker, but "unnatural" to a classical lover. The classical lover will think a more linear system sounds better, regardless of what music play through it. The same goes for the analog vs digital media argument. 137.71.23.54 17:05, 8 February 2007 (UTC)Brett

[edit] More rewrite needed

This page needs some significant re-writes. The stuff about the history is wrong in places - no mention of Sound In Syncs from the BBC in the late 60s. No mention of Nyquist except as a link at the bottom. Needs a separate section on the audio chain. Needs more examples of channel codes, and emphasis on the engineering benefits of digital systems, being careful not to imply that digital systems sound more/less pleasing than analogue systems. Needs some diagrams. Needs more of an engineering outlook.--Tomhannen 00:35, 7 February 2007 (UTC)

By all means: WP:BB. :) Cburnett 02:59, 7 February 2007 (UTC)

[edit] Hype

I love this medium, but:

The key advantage of digital audio is that the quality of audio signal being decoded is no longer directly dependant upon the quality of the recording medium. The ability to detect, and then to correct errors present in the recorded digital signal is abstracted to a mathematical problem instead of being a physical engineering problem. With the birth of cheap digital electronics, this has also made it easier and cheaper to copy and transmit digital signals.

This combines love of the abstruse with a subjective assessment. Let's just dive in to what digital audio *is*, first. --Uncle Ed 02:13, 25 February 2007 (UTC)

Hi. I added the first two sentences above. I think they are a good explanation of why digital audio has been widely adopted. I'd like to add them back to the main page, if you agree - I can live without the third one though... Is there anything fundamentally incorrect about the first two? I can find references for them in Watkinson - art of digital audio if that helps.. --Tomhannen 17:19, 25 February 2007 (UTC)

Then how about this:

Digital audio has been widely adopted because ...

Otherwise, no objections, and thanks for your effort and attention. --Uncle Ed 01:24, 26 February 2007 (UTC)

[edit] Hey everyone

Begun Restructure of digital audio introductory definition, use, and influence on the recording industry. I will continue this effort piece by piece when time allows. I plan to revise/rephrase, and aim for these additions to supplement some good content already written by others, any help, comments, suggestions, encouraged. I think the article was a little scattered and not structured well enough, but can be with a little effort from all of us.

I don't mean to sound cheesy, but it is what it is.

Later, James —The preceding unsigned comment was added by TheRecordingGuru (talk • contribs) 01:12, 20 July 2007 (UTC)

Hey James. I reverted your changes to this article as well, for some of the same reasons as on the recording studio article. Here's the link to your last-edited version. Some of my comments there apply here as well. One good resource to consult on and coordinate improvements would probably be the professional sound production WikiProject. I'll leave a welcome message on your user talk page with some other helpful links. Good luck! Dancter 01:45, 20 July 2007 (UTC)