Talk:Direct Stream Digital

Untitled
Has Direct Stream Digital Video ever been proposed? Photographic images of any kind, including those captured electronicly by TV/Video camaras, etc, are an analog phenomenen, similar to sound.
 * Yes. DSD is pulse-density modulation, a generalization of the pulse-width modulation used in Laserdisc. --Damian Yerrick (☎) 01:53, 1 February 2006 (UTC)

I just read about a format called DXD based on DSD. Sounds interesting, see link snippet below Christopher Sajdak 01:38, 29 December 2006 (UTC)

The advantages of DXD for SACD A new format, DXD (Digital eXtreme Definition for high quality and low noise recording and editing for SACD), has recently been acknowledged by Philips and Sony. DXD was initially developed for Merging’s Pyramix DSD workstation and recognised as one of the best formats for DSD source recording. Now that the first A-D and D-A convertors are available for converting direct into DXD, Digital Audio Denmark MD MIKAEL VEST gives a general overview of recording for SACD with an outline of the pros and cons of DXD.

Does anyone think that the article needs to cover more of the noise-shaping algorithms that DSD recording devices use? I am not the person to ask regarding this subject but I think it's an important difference between PCM and DSD. Ptmoore 22:50, 7 May 2007 (UTC)

Sentence is confusing; also, a terrible way to describe DSD
This sentence is ambiguous: "The process of creating a DSD signal is conceptually similar to taking a 1-bit delta-sigma analog-to-digital (A/D) converter and removing the decimator, which converts the 1-bit bitstream into multibit PCM." It is also a terrible way to describe the process in an introductory article. It should be described in a simpler fashion, from basic principles, instead of relying on the reader first to understand a more complicated system. — Preceding unsigned comment added by Oscarruitt (talk • contribs) 06:46, 23 June 2011 (UTC)


 * A better description would be this: "The 1-bit sigma-delta converter keeps a running total of the proceeding single bit DSD stream (the sigma-part), where a zero is -1 and a one is +1, and compares this is to a sample of the input signal (at, say, 24-bit resolution using a analogue-to-digital converter or ADC) (the delta-part) to calculate whether the next bit should be a +1 (if the input is greater than sum) or a -1 (if the input signal is less than the sum). If the sum is the same as the input signal the converter still has to choose +1 or -1, creating noise while "hunting" for the precise value. If the input signal is sampled at 96 kHz, for example, there are 29.4 DSD samples (at 2.8224 MHz for SACD) for every PCM sample." Derekjc (talk) 12:01, 1 June 2013 (UTC)

Figure is wrong
The first figure, the one showing waveforms for PCM and DSD, is just wrong and horrible and should be removed immediately. It looks like it was drawn by a child, a child who doesn't understand sampling. Someone remove it now. — Preceding unsigned comment added by Oscarruitt (talk • contribs) 05:46, 23 June 2011 (UTC)

quality
when i run spectral analysis on pcm wav ripped from CDs, many times the frequency rate peaks below 44.1, also consider the generations of recording on dats. have any recordings actually utilized the extremely high frequency possible with this format? i'd like to have moar information on this subject plz. -- Alex Ov  Shaolin  01:08, 4 August 2007 (UTC)


 * This is a fairly standard thing - studios often run 20khz low pass filters on their final mixes to avoid any ringing artefacts on CD from frequencies near/over the 22.05khz Nyquist limit of the 44.1khz media, and probably to reduce treble noise on analogue media (don't think I've come across a tape out of any I've analysed that mustered 20khz, regardless of what was prerecorded on it, or what the sales pitch on the shrinkwrap of blank cassette may have said). A number of them do go pretty much all the way up to 22khz, the only suggestion I can make for these differences is the age of the recording - newer equipment will have better quality and largely digital filters that can make a very effective, narrow 'cut', say with the filter output dropping sharply from 0db at 21.5khz to -96dB at 22.0khz, but older mixers will have much looser, cheaper, analogue filters that might need a good couple khz of bandwidth to muster a -60dB drop, and so will start rolling off from around 19khz in order to knock out enough of the high frequencies. Also any discs made from older material will suffer similar problems simply from lower fidelity master material, when it didn't matter much to preserve the very highest frequencies as vinyl couldn't easily represent them even on the first play, and rapidly lost treble fidelity on repeat plays. (I have no citations, but a lot of this knowledge came from wikipedia and pages linked from it - do a search :) 82.46.180.56 (talk) 02:35, 27 December 2007 (UTC)

Fair use rationale for Image:DSDlogo.gif
Image:DSDlogo.gif is being used on this article. I notice the image page specifies that the image is being used under fair use but there is no explanation or rationale as to why its use in this Wikipedia article constitutes fair use. In addition to the boilerplate fair use template, you must also write out on the image description page a specific explanation or rationale for why using this image in each article is consistent with fair use.

Please go to the image description page and edit it to include a fair use rationale. Using one of the templates at Fair use rationale guideline is an easy way to insure that your image is in compliance with Wikipedia policy, but remember that you must complete the template. Do not simply insert a blank template on an image page.

If there is other fair use media, consider checking that you have specified the fair use rationale on the other images used on this page. Note that any fair use images lacking such an explanation can be deleted one week after being tagged, as described on criteria for speedy deletion. If you have any questions please ask them at the Media copyright questions page. Thank you.

BetacommandBot (talk) 06:36, 2 January 2008 (UTC)

DSD64, DSD128, and DXD
These variants of DSD are of interest and should be added by someone who knows about them much better than me. —Preceding unsigned comment added by 78.105.11.183 (talk) 00:28, 11 April 2008 (UTC)
 * DSD128 is now added (DSD64 is regular DSD) while DXD has its own article (to which a reference is now added). The Seventh Taylor (talk) 23:10, 9 March 2009 (UTC)

Proposal to merge DSD-CD article with DSD article
As original contributor of the DSD-CD article I do not object to merging it with this article as suggested. I even volunteer doing it myself. The main reason for making it a separate article was to be able to include a link to it in Template:Audio formats but I suppose that can be achieved otherwise too. The Seventh Taylor (talk) 22:27, 30 May 2009 (UTC)

DSD/PCM explanation image
I'm fairly sure that the picture explaining the difference between PCM and PDM is incorrect, as it shows the (admittedly similar) PWM signal, not PDM. In PWM the length of the pulse denotes amplitude, as in the image. However in PDM, as the pulses are not contiguous, so it would appear to the casual observer more like higher frequency. Example: In PWM (NOT DSD) we can denote the lowest amplitude as 0000, slightly higher signal amplitude as 0001, then 0011, then 0111 and the highest amplitude as 1111 (not to be confused with binary of course), so a continuous low amplitude would be 0000/0000/0000, and a rising signal would be represented as 0000/0001/0011/0111/1111. Each slash-parenthesis represents one sample. In contrast, PDM, as used in DSD, should be (in ascending order) 0000/0001/0101/1111. Note additionally that there is less information transmitted with a given master clock rate. Can somebody either change or replace the photo? It's great having images, people like me can't figure stuff out without 'em :) Arctic hobo (talk) 21:41, 25 September 2009 (UTC)

-Also, have only just realised that the image for PCM is misleading as well - it shows PCM as a simple quantisation of the analogue signal - but of course PCM, PDM and PWM all quantise analogue signals, that's what they're for! An image showing the parallel nature of PCM (ie it is transmitted as binary words) with several simulataneous pulses corresponding to the binary code for amplitude is needed. I'm pretty new to wikipedia editing but can someone delete or modify the image? Cheers 77.103.195.66 (talk) 14:27, 26 September 2009 (UTC)

Let me throw my oar in, as well. It just doesn't look right.

I'm confused in another way. I assumed DSD worked simply by comparing sample (n+1) to sample (n). If the former were equal to or larger than the latter, a 1 was transmitted; if not, a zero. Such a signal could theoretically be converted to analog simply by running it through a low-pass filter (as the article states).

This is very much a "hand-waving" article that needs a lot more information about exactly how DSD works. This lack of clear, easily followed explanation is a major failing of many Wikipedia technical articles. WilliamSommerwerck (talk) 16:14, 7 February 2011 (UTC)

Unwarranted conclusion from Detmold study?
The stated conclusion does not follow from the premise: it may be true that "hardly any of the subjects could make a reproducible distinction between the two encoding systems". However, if even one subject could make a reliable distinction, the conclusion that "no significant differences are audible" does not follow.

I remember a similar study wherein listeners were tested to see if they could distinguish beween a Strad and a top modern violin. The vast majority could not. However, a few could tell the Strad every time.

The Detmold paper seems to contain nothing about the age of the listeners, either, which makes me wonder how meaningful it can be. Paul Magnussen (talk) 20:47, 21 December 2009 (UTC)
 * You know, the science shows that people can not hear the difference between 16/44.1 (CD quality) and higher bitrates. The most famous recent study showing this is the Meyer/Moran paper Audibility of a CD-Standard A/D/A Loop Inserted into High-Resolution Audio Playback, but there’s also Toshiyuki Nishiguchi, Kimio Hamasaki, Masakazu Iwaki, and Akio Ando, "Perceptual Discrimination between Musical Sounds with and without Very High Frequency Components", and, yeah, the people who make lossy audio compression codecs do a lot, and I mean, a lot of double blind tests and are sure you can't hear the difference between 16/44.1 and a wire.
 * There are people who desperately want to believe that maybe people (some subset of the population, maybe) can really hear the difference between 16/44.1 and something else with higher resolution, but every scientific study I've seen shows otherwise, with the possible exception of hypersonic effects where frequencies this high affect our subconscious perception of music.
 * In terms of Wikipedia policy, since a number of peer-reviewed papers show 16/44.1 sounds as good as a wire, and there isn’t a single scientific paper showing it doesn’t, we need to reflect what reliable sources say about the subject. Also, it is not a good idea to add original research criticizing scientific papers; if the papers are flawed, we need to find other peer-reviewed papers criticizing them.
 * If there really is something we're not hearing that is affecting how we experience music, why stop at 384/24 PCM or 2822.4/1 DSD; let's store a 4-bit 6.144 MHz signal (that's 64x 96khz, guys). To store this signal, uncompressed, requires 3,072,000 bytes per second per channel (stereo is two channels. Compare this to the 88,200 bytes per second 16/44.1 needs, the 352,800 bytes per second DSD needs, and the 576,000 bytes per second 24/192 needs.), the internal rate of a more modern converter than what the DSD converters use. Samboy (talk) 07:34, 22 December 2009 (UTC)

There's a related issue that can't be discussed in the article, but needs to be brought up here. I'm a degreed EE who doesn't believe that numbers -- or even double-blind tests -- tell us everything. It's not that they can't -- any audible difference must be measurable -- it's that no one has done the research that would create a useful correlation -- probably because it's against both sides' interest to have any genuinely objective way to measure audio equipment. In particular, those inveighing against SACD have a personal motive in doing so -- many of them supported the 44.1/16 as delivering "pure, perfect sound". If SACD is subjectively more accurate, then their Weltanschauung goes right out the finestra. As for me... I have hundreds of SACDs. Isn't it amazing I can tolerate listening to such awful sound? WilliamSommerwerck (talk) 16:26, 7 February 2011 (UTC)

DSD vs. PCM
The first paragraph of the DSD vs. PCM had a long explanation with no citations and at least a few fundamental inaccuracies.(eg: it states that sound is a transverse wave and light is longitudinal, in reality they are reversed)

I'm removing this until it is reviewed, preferably by the original author. — Preceding unsigned comment added by 2601:A:1700:6700:0:0:0:E96 (talk) 13:59, 16 May 2015 (UTC)

I appreciate seeing that my erroneous edit was removed. I apologize for getting those waves reversed. I do feel it is important to discuss the difference in the waves as part of any comparison of PCM and DSD, although I am clearly not the most qualified to do so. I look forward to seeing an edit with this information from a qualified contributor to this topic. Thank you. CareerPilot1971 (talk) 06:32, 21 May 2015 (UTC)

Who is Stuart? He is mentioned without introduction.

External links modified
Hello fellow Wikipedians,

I have just added archive links to 2 one external links on Direct Stream Digital. Please take a moment to review my edit. If necessary, add after the link to keep me from modifying it. Alternatively, you can add to keep me off the page altogether. I made the following changes:
 * Added archive https://web.archive.org/20060823173606/http://www.acoustics.salford.ac.uk:80/research/angus_files/angus_files/publications/publications.htm to http://www.acoustics.salford.ac.uk/research/angus_files/angus_files/publications/publications.htm
 * Added archive https://web.archive.org/20061007044755/http://www.diamondcenter.net:80/digitalstress.html to http://www.diamondcenter.net/digitalstress.html

When you have finished reviewing my changes, please set the checked parameter below to true to let others know.

Cheers. —cyberbot II  Talk to my owner :Online 04:19, 18 October 2015 (UTC)

External links modified
Hello fellow Wikipedians,

I have just added archive links to 2 one external links on Direct Stream Digital. Please take a moment to review my edit. If necessary, add after the link to keep me from modifying it. Alternatively, you can add to keep me off the page altogether. I made the following changes:
 * Added archive http://web.archive.org/web/20100602073553/http://www.overture.org.ua:80/files/SDPC.pdf to http://www.overture.org.ua/files/SDPC.pdf
 * Added archive http://web.archive.org/web/20060518133100/http://www.acoustics.salford.ac.uk:80/research/angus_files/angus_files/publications/publications.htm to http://www.acoustics.salford.ac.uk/research/angus_files/angus_files/publications/publications.htm

When you have finished reviewing my changes, please set the checked parameter below to true or failed to let others know (documentation at Sourcecheck).

Cheers.—cyberbot II  Talk to my owner :Online 06:04, 28 February 2016 (UTC)

Flagged DSD technique section for incomprehensible, abbreviations, and technicality
I flagged that section because after several attempts I was simply not able to sift through the audio equipment jargon and abbreviations to understand how DSD works. And I generally understand how audio sampling works. Someone who knows more about DSD would do well to clarify what is being said with less jargon and clearer explanations of the mechanism of DSD. There are links to related concepts but even after looking at those I still don't really know what's being said. For example this bit:

"The process of creating a DSD signal is conceptually similar to taking a one-bit delta-sigma analog-to-digital (A/D) converter and removing the decimator, which converts the 1-bit bitstream into multi-bit PCM. Instead, the 1-bit signal is recorded directly and, in theory, only requires a lowpass filter to reconstruct the original analog waveform. In reality it is a little more complex, and the analogy is incomplete in that 1-bit sigma-delta converters are these days rather unusual, one reason being that a one-bit signal cannot be dithered properly: most modern sigma-delta converters are multi-bit."

Creating a DSD signal is "conceptually similar" to [jargon jargon jargon], which [jargon] but [qualification with jargon] and also the concept isn't really simmilar because [jargon].

Dcb2124 (talk) 16:09, 23 March 2016 (UTC)

After reading the section, I agree with you. It is very hard to understand this section, even though I understand how DSD works! Here is a concept I think should be included in this section. The DSD signal is sampled at 1 bit, so it's either being sampled as a +1 or a -1 from the previous sample. One of the problems is that we don't know from the native DSD signal how much the signal moved, only that it moved up or down.

So a fast sampling rate is necessary for DSD to work. Also the Nyquist-Shannon sampling theorem isn't directly applicable to DSD but rather it's the Poisson special case of the theorem. On the other hand the theorem also gives us that as the bit depth and sampling rate of PCM encoding tends to infinity, this is equivalent to the sampling rate going to infinity for DSD and increasing headroom for both. Therefore the imperfections in implementation that will be different from the theory is that the theorem requires infinite bit depth for PCM and infinite sampling rate for DSD.

At the limits of PCM sampling rates, and for any natural sound, the difference in amplitude of adjacent samples is also going to tend to be infinitely small, basically either bigger in amplitude than the previous signal, the same amplitude, or less amplitude. Here we see the DSD problem, it being 1 bit cannot encode the signal staying the same. Thus to encode the signal staying the same it goes up and down very fast in amplitude in the DSD encoding, or just a series of alternating bits of 0101010 ... This is the dithering aspect of the DSD, which is employed in practice and theoretically can be described by the Shannon theorem special case as a specific low pass filter.

Then, as you increase in bit depth in DSD, even if it doesn't encode for 0 signal change the distortion introduced by dithering can be reduced, but this is going back to PCM-type sampling then.

Another extreme case is when a series increases in amplitude is recorded by ordinary DSD, it can therefore encode incredibly loud low frequency sounds.

It should be noted that in theoretical settings, with a countably infinite sampling rate and bit depth, DSD is the first derivative of the PCM encoding.

87.116.161.79 (talk) 18:39, 9 May 2021 (UTC)

External links modified
Hello fellow Wikipedians,

I have just modified 4 one external links on Direct Stream Digital. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Corrected formatting/usage for http://www.dmprecords.com/index.html
 * Added archive https://web.archive.org/web/20131203021355/http://mixonline.com/mag/audio_recording_vibes_ese/index.html to http://mixonline.com/mag/audio_recording_vibes_ese/index.html
 * Corrected formatting/usage for http://dsd-guide.com/usb-link-dsd-audio-pcm-frames-andreas-koch
 * Corrected formatting/usage for http://www.smr-home-theatre.org/surround2002/technology/page_07.shtml

When you have finished reviewing my changes, please set the checked parameter below to true or failed to let others know (documentation at ).

Cheers.— InternetArchiveBot  (Report bug) 07:16, 9 November 2016 (UTC)

External links modified
Hello fellow Wikipedians,

I have just modified 2 one external links on Direct Stream Digital. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Added tag to http://www.marketnews.ca/LatestNewsHeadlines/AcousticSoundsLaunchesHi-ResDownloadService.html
 * Added archive https://web.archive.org/web/20060513223001/http://www.essex.ac.uk:80/ese/research/audio_lab/malcolmspubdocs/C115%20SDM%20versus%20LPCM.pdf to http://www.essex.ac.uk/ESE/research/audio_lab/malcolmspubdocs/C115%20SDM%20versus%20LPCM.pdf
 * Added archive https://web.archive.org/web/20070927200955/http://www.hfm-detmold.de/eti/projekte/diplomarbeiten/dsdvspcm/aes_paper_6086.pdf to http://www.hfm-detmold.de/eti/projekte/diplomarbeiten/dsdvspcm/aes_paper_6086.pdf

When you have finished reviewing my changes, please set the checked parameter below to true or failed to let others know (documentation at ).

Cheers.— InternetArchiveBot  (Report bug) 16:31, 13 December 2016 (UTC)

External links modified
Hello fellow Wikipedians,

I have just modified 4 external links on Direct Stream Digital. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Added archive https://web.archive.org/web/20120403041109/http://www.mofi.com/store/pc/showsearchresults.asp?pageStyle=P&resultCnt=10&keyword=sacd&submit=Go+%3E%3E to http://www.mofi.com/store/pc/showsearchresults.asp?pageStyle=P&resultCnt=10&keyword=sacd&submit=Go+%3E%3E
 * Corrected formatting/usage for http://dsd-guide.com/usb-link-dsd-audio-pcm-frames-andreas-koch
 * Added archive https://web.archive.org/web/20090303232815/http://www.diyaudio.org.ua/download/file.php?id=242 to http://www.diyaudio.org.ua/download/file.php?id=242
 * Added archive https://web.archive.org/web/20080901162000/http://www.dsd-format.sony.net/en/index.php to http://www.dsd-format.sony.net/en/index.php

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

Cheers.— InternetArchiveBot  (Report bug) 19:39, 3 September 2017 (UTC)

Resolution Reduction Calculation is Wrong
The second paragraph states that the "sequence of single-bit values ... [is stored] ... only at $1⁄32768$ of its 16-bit resolution)."

This seems incorrrect. A resolution reduction from 16 bits to 1 bit is 2-16, so the correct reduction should be $1⁄65536$. If the reduction is from a stereo pair to the single-bit stream (???), this number would be $1⁄131072$.

algocu (talk) 14:32, 18 November 2017 (UTC)


 * Thought so as well. But the sentence talks about a "reduction": From 16bits to 1bit is a reduction of 15 bits, which is equivalent to 2^-15. 212.35.11.31 (talk) 12:20, 11 September 2022 (UTC)

Lead improvements
I have reverted lead changes. These are not clear improvements and the edit summary does not help me understand what improvements are intended. ~Kvng (talk) 22:02, 13 January 2023 (UTC)

DSD64 is NOT 64x a CD
DSD64 sampling is NOT 64x that of a CD. It’s DOUBLE. CD is 16-bit/44.1kHz. Multiplied together to get the total number of bps, it’s 705.6kbps PER CHANNEL. Comes out to 1,411.2kbps total. That’s 1.4112MHz. DSD64 is 2.8224MHz which is EXACTLY DOUBLE because it comes to the same 16-bit depth and twice the sampling frequency of 88.2kHz. DSD64 advertising that it samples twice as much as CD is FALSE. This information needs to be updated! 2600:1700:B291:F30:E5EA:53F7:F0A1:FEE8 (talk) 05:51, 26 August 2023 (UTC)


 * The sample rate (the rate at which the audio signal is examined) is 64x. The information rate is 2x. ~Kvng (talk) 14:19, 28 August 2023 (UTC)

SACD cannot have 80-90kHz response
If SACD is DSD64, it cannot have a frequency response of 80-90kHz! It’s the equivalent of 16-bit/88.2kHz. It has to sample a wave TWICE to be able to see a wave. That gives it a maximum frequency response of 44.1kHz. There are MANY errors in this article and it needs to be rewritten. 2600:1700:B291:F30:E5EA:53F7:F0A1:FEE8 (talk) 05:59, 26 August 2023 (UTC)


 * Nyquist–Shannon sampling theorem says it has a maximum frequency response of 1.4112 MHz. Of course noise shaping buries most of that so determining frequency response is somewhat subjective. ~Kvng (talk) 14:25, 28 August 2023 (UTC)