Talk:Sampling rate

From Wikipedia, the free encyclopedia
Jump to: navigation, search
WikiProject Professional sound production (Rated Start-class, Mid-importance)
WikiProject icon This redirect is within the scope of WikiProject Professional sound production, a collaborative effort to improve the coverage of sound recording and reproduction on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
Start-Class article Start  This redirect does not require a rating on the project's quality scale.
 Mid  This redirect has been rated as Mid-importance on the project's importance scale.

Name of article[edit]

Question: should this be at Sampling frequency or Sampling rate? The latter seems to be linked a bit more often. (Or is there some obscure difference between the meaning of the two terms that I'm not aware of?) --Brion

Google reveals far more hits for "sampling rate" than "sampling frequency", so I have moved the page there per Wikipedia:Use common names. —Lowellian (reply) 09:25, 7 July 2006 (UTC)

The title should stay as "Sampling rate". The word frequency is reserved for a specific parameter of the sinusoid function. There is nothing sinusoidal about sampling. Further to this point: Sampling rate should never be expressed in Hz. This is a common error. Sampling rate should be expressed in samples per second (sps). —Preceding unsigned comment added by (talk) 23:23, 9 March 2009 (UTC)


Regarding video: the sampling frequency is the frame/field rate, rather than the notional pixel clock. All modern TV cameras use CCDs, and the image sampling frequency is the repetition rate of the CCD integration period. -- Anon.

—Preceding unsigned comment added by (talk) 16:57, March 18, 2003

"All modern TV cameras use CCDs" - Not true: some modern TV cameras use CMOS sensors. Ikegami's new HD cameras come to mind. Ehusman 23:40, 8 April 2006 (UTC)



96,000 or 192,400 Hz - DVD, BD-ROM (Blu-ray Disc), and HD-DVD (High-Definition DVD) Audio

They seriously are doing 192.4 KHz sampling rate for audio? Or is 192 two-channel 96 KHz or something? I seriously can't believe they would waste that much space on audio considering that's many times higher than our hearing can hear. Cburnett 04:35, 25 Feb 2005 (UTC)

You may be surprized, but many professional recorders digitize sound at 192kHz. Remember that recording at a higher resolution will better sample the soundwave and therefore less interpolation is required to recreat the sound when you play it back. A standard cd can only play back sound up to 20,500 Hz in theory, but if you have sound at 20,000 Hz you only have two samples representing the sound wave. Hopefully you can appreciate that 192kHz, although capable of reproducing every high pitched sounds, is used to produce more accurate sound in the range of human hearing. -- 21:34, 29 November 2005 (UTC)
A good rule of thumb for high fidelity sampling is to sample 7-10 higher than the highest frequency of interest.--User:Ehusman 23:14, 30 November 2005 (UTC)

192.4 kHz is wrong, it should be even 192.0 kHz. I guess someone has mixed this up with 176.4 kHz.

Audio vs. video sampling rates[edit]

Question: Usually, voice is sampled at 8000 samples per second, but video may be sampled at 6000000 samples per second. Why do the two types of signal require different sampling rates? —Preceding unsigned comment added by (talk) 04:20, April 3, 2006

Human hearing operates between 20 Hz and 20 kHz, so the human voice operates well within these two extremes. The human eye only gets "refreshed" about 24 times per second; anything that happens more frequently than that gets blurred together (known as "persistence"). You must sample at a frequency twice the highest frequency you wish to store (something called Nyquist's sampling theorem). Thus, sound is sampled at relatively low rates (you can probably still preserve most voices at 8000 Hz), but video is only at 60 Hz. Your figure of 6000000 Hz is way off - unless you are counting individual pixels or ultra high speed photography, which is a specialized field. Ehusman 23:40, 8 April 2006 (UTC)

1.001 Hz or 1,001 Hz ??[edit]

The "60 / 1.001 Hz - NTSC video" under "Video systems", should it be "60 / 1,001 Hz" ?? Is it thousand and zero zero one Hz or one dot zero zero one Hz ?? —The preceding unsigned comment was added by (talk) 10:18, 9 May 2007 (UTC).

This article could be improved[edit]

I looked up this article to find out the sampling rate for audio CD, DVD audio, studio quality audio etc., but could not find it. --Andreas Rejbrand (talk) 20:57, 15 May 2009 (UTC)

I looked up this article to find out why most of my MP3s say Sample Rate 44.100 kHz, but a few say Sample Rate 48.000 kHz. I was looking for information written in layman's language. Did I miss it? (talk) 16:45, 2 July 2010 (UTC) Regards, Jim.

Problem with introduction[edit]

Yes, there is one particular problem about the introduction of this article, namely that sample (for which you seem to use the letter "S") is not a physical unit according to SI. Are you sure that RF Design Magazine is a so-called reliable source? Akilaa (talk) 08:14, 25 December 2009 (UTC)

Here are a few hundred books that could be used as alternative sources. Dicklyon (talk) 17:21, 25 December 2009 (UTC)

Any number of sources is not important if you don't yourself understand this concept. I'm sorry, but you answered only to the less important question. The main point was that sample is not a physical unit because they can only be measured with integers but not real numbers. Only when a measure can be measured with real numbers, then a unit is physical. Otherwise 1/x shall be used. Even though you have such a nice gallery, it doesn't mean that you can start to claim anything. Akilaa (talk) 18:18, 27 December 2009 (UTC)


This article contains contradictory statements:

"The Nyquist–Shannon sampling theorem states that perfect reconstruction of a signal is possible when the sampling frequency is greater than twice the maximum frequency of the signal being sampled"


"It is a common error to interpret Nyquist theorem that with a sampling frequency of twice that of the original signal is enough for perfect reconstruction."

I suspect that the first passage is in need of some sort of qualifier, but I do not know enough of the subject to fix this. Tomyhoi (talk) 16:59, 27 February 2010 (UTC)

---added On the same topic, I was told in my control classes that you want a sampling rate of 4-5 times the max frequency and that you need to sample at twice the max frequency just to be able to determine the frequency of the source. And then at 2x hz, you can't discriminate between a sine wave, square wave or saw wave. (mjb 7May10) —Preceding unsigned comment added by (talk) 02:08, 8 May 2010 (UTC)

what is MS/s?[edit]

Ms/s or MS/s redirects here, but the current page doesn't talk about that . I understand it is mega samples per second, but this should be included in the text (maybe briefly referring to the debate , included above, over whether this is a correct measurement or not). I mean, National Instruments use MS/S so it is surely in common use. —Preceding unsigned comment added by Dirkjot (talkcontribs) 11:51, 27 May 2010 (UTC)

Merge to Sampling (signal processing)[edit]

A lot of the material in this article is repeated over at Sampling (signal processing). Basic information about sampling does belong there and should be removed or merged from here. Does that leave enough of an article here? Should this whole article be merged into Sampling (signal processing)? --Kvng (talk) 19:20, 10 September 2010 (UTC)

I see no reason to keep a separate article. If you can do a decent merge that doesn't lose anything, and make a section on sampling rate in the sampling article, then I don't imagine anyone would object to making this article a redirect. But then sometimes I do get surprised... Dicklyon (talk) 22:58, 10 September 2010 (UTC)
I agree that both articles should be merged. — Preceding unsigned comment added by (talk) 19:29, 1 October 2013 (UTC)
I think I also approve of the merge though I don't think the list of sample rates belongs under Sampling (signal processing). Maybe a List of common sample rates should be created or the list removed. Radiodef (talk) 02:27, 23 November 2013 (UTC)
I merged this article's info into that one. But I stopped short of changing this one into a redirect page, because it would create a lot of double redirects, and I'm not sure which ones have to be fixed manually and which ones are fixed automatically, by robots.
--Bob K (talk) 02:35, 14 March 2014 (UTC)


This article needs audio examples, so we can hear the difference.--RaptorHunter (talk) 05:56, 15 November 2010 (UTC)

Proposed examples would be material low-pass filtered at 1/2 sampling frequency. In many cases the difference would not be obvious. --Kvng (talk) 15:19, 15 November 2010 (UTC)

Unexplained Notation[edit]

The second paragraph currently reads, "Sample rate is usually noted in Sa/s (non-SI) and expanded as kSa/s, MSa/s, etc. The common notation for sampling frequency is fs which stands for frequency (subscript) sampled."

  • What does "Sa/s" mean?
  • What does "kSa/s" mean?
  • What does "MSa/s" mean?
  • Why is "fs" a "notation" and not an "abbreviation"?

I would fix this if it was abundantly clear what it stood for. --Kebman (talk) 08:43, 23 June 2011 (UTC)

It was a bit confusing, and I merged parts of the paragraph to where it should make sense. Basically, "Sa/s" is "samples per second", and the "k" and "M" are SI prefixes for ×103 and ×106, respectively. However, I don't think that this article really needs to explain all of the SI prefixes, so I didn't keep or elaborate them (as there are potentially many of these). "Sa/s" is not very common and is essentially the same as 1/s or Hz. Hope it is a bit more clear. +mt 10:37, 23 June 2011 (UTC)

Time centrism intro[edit]

Technically, you could be sampling something that is a function of some variable other than time, like a scanner's resolution of "1200 dpi" is also a sampling rate, or you could be doing image processing at arbitrary pixel spacings. Time and hertz are common and easy to understand, but maybe throw in some mentions of other applications. — Preceding unsigned comment added by Justanothervisitor (talkcontribs) 17:00, 26 November 2012 (UTC)

The intro to Sampling (signal processing) does not have this issue. Another small benefit of doing the proposed merge. ~KvnG 21:51, 3 October 2013 (UTC)

Misleading and contradictory in progression of explaining scale[edit]

Coming at this article and topic somewhat anew, the progression of the article makes little sense. The first section makes it dead clear that 44.1 kHz is the absolute sampling limit of the known universe. The 2nd P.S's that if you really want, you can overdo things if you have a compulsion for a "less steep analog anti-aliasing filter". Only when you read the entire "more complete" list do you have any general idea that higher and higher sampling rates are associated with more and more professional equipment and audio systems. Why would I ever want 5,644,800 Hz if I already can get "perfect reconstruction" at already double my hearing range of "20 kHz", with a simple compact disc? I don't still get this after reading the article, I just see fancy million-dollar stuff at the high end of the list...

It's an informative article but the progress makes no sense. Squish7 (talk) 06:40, 4 February 2014 (UTC)

I'm not sure I see the problem. After a definition in the lead, and general sections on the sampling theorem and on the possibilities of over and under sampling, we have a section on audio. What would you suggest? Dicklyon (talk) 06:46, 4 February 2014 (UTC)
As to the question of why anyone would ever prefer to use 5,644,800 Hz sample rate, I think that's mostly a marketing thing; bigger numbers are better. Certainly the sampling theorem should convince one that high quality is quite possible at much lower sample rates, like with CDs at 44.1 kHz. Dicklyon (talk) 06:53, 4 February 2014 (UTC)
That makes no sense. Why would DVD and Blu-ray sample up to 192 kHz when 44 kHz is the maximum we can hear? This is all just scam talk? Squish7 (talk) 07:03, 4 February 2014 (UTC)
Perhaps this excerpt from Nyquist–Shannon sampling theorem sheds some light on the question:
  • From Figure 5, it is clear that larger-than-necessary values of fs (smaller values of T), called oversampling, have no effect on the outcome of the reconstruction and have the benefit of leaving room for a transition band in which H(f) is free to take intermediate values. Undersampling, which causes aliasing, is not in general a reversible operation.
  • Theoretically, the interpolation formula can be implemented as a low pass filter, whose impulse response is sinc(t/T) and whose input is \textstyle\sum_{n=-\infty}^{\infty} x(nT)\cdot \delta(t - nT), which is a Dirac comb function modulated by the signal samples. Practical digital-to-analog converters (DAC) implement an approximation like the zero-order hold. In that case, oversampling can reduce the approximation error.
--Bob K (talk) 13:15, 13 March 2014 (UTC)