Talk:Sampling (signal processing)

Some reorganization
I propose to merge Discrete-time signal into this article and also the material in Digital_signal that was recently split from Discrete-time signal by. This will put description of the signal that results from sampling in context of the process that creates it. Covering Discrete-time signal comprehensively will result in a lot of unnecessary overlap with this article (Sampling (signal processing)). Removing Digital_signal from that article will allow Digital signal to be about one topic - the pulse train signals used for digital logic. Digital signal is now part disambiguation and part article which is not good. ~Kvng (talk) 15:14, 25 July 2015 (UTC)


 * I would oppose such a merge. There are other ways to get discrete-time signals than by sampling continuous-time signals.  These are distinct topics.  73.158.249.163 (talk) 03:16, 26 July 2015 (UTC)


 * Can someone give me an example or two of the other ways to get discrete-time signals. Discrete-time signal lists two but they are both sampling. ~Kvng (talk) 13:18, 28 July 2015 (UTC)


 * I agree with relocating Digital_signal to allow that article to focus on digital logic. But I also prefer to keep it physically separate from this article, connected by a WikiLink.  A more appropriate location is Quantization (signal processing), and I have attempted that merge.  But I have not yet removed anything from Digital_signal.
 * Discrete-time signal is more like a handy definition than an article, which is very useful (see WhatLinksHere). I think that list is sufficient justification for the article's independence, and I don't think it needs to grow and cover the topic comprehensively, because it can WikiLink back here (and elsewhere) for that.  Also, the Discrete-time_signal section, which makes a valid distinction between sampling a continuous function vs observing an inherently discrete process, doesn't really need to be here.  I think it is better off encapsulated where it is.
 * --Bob K (talk) 17:46, 29 July 2015 (UTC)


 * Do you have a suggested destination for relocating Digital_signal?
 * I've done a bit of cleanup in this topic area and discovered another overlapping article: Discrete time and continuous time. Any opinions about this material? ~Kvng (talk) 21:01, 29 July 2015 (UTC)


 * My suggestion was Quantization (signal processing), and (as I said above) I have already done it . It is not just a cut & paste merge.  It is a blend, in the interest of coherence.
 * I just took a quick look at Discrete time and continuous time. IMO, it's not very interesting.  I would probably support a movement to delete it.  But if others find it useful, I'm happy to just leave it alone.
 * --Bob K (talk) 16:12, 30 July 2015 (UTC)


 * You have done some work on Quantization (signal processing) but Digital signal has not been changed. ~Kvng (talk) 16:58, 1 August 2015 (UTC)

A related topic is that the What Links Here list for Digital_signal has 309 entries, and oddly, Digital Signal Processing is not one of them. Perhaps the name "Digital Signal" is a misnomer. Would "Boolean signal" or "Data signal" be more appropriate? It is hard to tell what those 309 links are trying to reach, because Digital_signal is about two very different topics. If we purify it, by removing the sampling/quantizing parts, how many of those 309 links will lose the content they wanted? --Bob K (talk) 15:54, 1 August 2015 (UTC)


 * I think it is best to move and continue this discussion at Talk:Digital_signal ~Kvng (talk) 16:58, 1 August 2015 (UTC)

Hint for Audio Studio, Nyquist-Theorem
44.1 kHz is considered fine (per timed sample) enough, to get an exact impression in the human ear, but:

If you got several digital instruments, and their output DACs use a 44.1 kHz Sampling Rate, i would suggest a 88.2kHz recording samplerate. See Nyquist Theorem (Sampling Frequency).

If you got high quality DACs and ADCs, the same sampling rate could give good results too, even if the converters work unsynced (because the energy potential flows "analog").

MfG, Hannes E. Schäuble

PS: Yes, sometimes you hear the difference between 44.1 and 88.2 kHz ;) e.g. "natural" overtones etc.

And yes, some musicians do use frequencies you can't head with your ear, but bones etc. intentionally. Sad if they are lost. (e.g. Sub-Bass in old Church Pipe Organs)

MS/s
I was redirected from a page named "MS/s". I think it means Mega-Samples per second, but want to confirm. Unfortunately, the units S/s does not appear anywhere on the page. Anyone know the answer? — Preceding unsigned comment added by 2400:DD01:2000:15:18F4:FB1D:F641:5A3F (talk) 08:06, 4 July 2016 (UTC)


 * I added it to the page. Correct if wrong! — Preceding unsigned comment added by 2400:DD01:2000:15:18F4:FB1D:F641:5A3F (talk) 04:38, 5 July 2016 (UTC)

Foldback aliasing
The article talks about foldback aliasing in the sampling rate section (3.1.1). I know what is aliasing, but I've never heard this term, nor can I find any reference to it on the internet. — Preceding unsigned comment added by Rossengers (talk • contribs) 02:59, 12 July 2019 (UTC)
 * This internet thing is a loss. Look in books. Dicklyon (talk) 03:24, 12 July 2019 (UTC)

Sampling Rate and Human Hearing Range
"When it is necessary to capture audio covering the entire 20–20,000 Hz range of human hearing,[5] such as when recording music or many types of acoustic events, audio waveforms are typically sampled at 44.1 kHz (CD), 48 kHz, 88.2 kHz, or 96 kHz."

I don't see what the sampling rate has to do with the range of human hearing. They have nothing to do with each other. Sampling is done in order to preserve and convert a signal from one format to another. I think this line here is incredibly misleading and gives the impression that sampling has something to do with being able to hear the full range of frequencies available instead of taking discrete samples of a waveform. Quite simply it is about accurate reproduction of a signal. Keyword being accuracy

"Sampling rates higher than about 50 kHz to 60 kHz cannot supply more usable information for human listeners"

Completely uncited, followed by this:

"There has been an industry trend towards sampling rates well beyond the basic requirements: such as 96 kHz and even 192 kHz"

Wow I'd really like to know if there's more to that, wish Wikipedia was still an encyclopedia worth a damn.

"Even though ultrasonic frequencies are inaudible to humans"

Once again, sampling has nothing to do with being able to hear greater than 20kHz or less than 20Hz. This entire section should ideally be rewritten as it's promoting this myth that sampling rate is somehow linked to hearing range or hearing a greater range of frequencies in audio. 86.41.241.197 (talk) 13:02, 23 November 2019 (UTC)


 * I hear you! (pun intended). The white noise page has this same incorrect information. Embarassing, isn't it. — Preceding unsigned comment added by 203.166.252.152 (talk) 14:59, 18 January 2024 (UTC)

I assume you're just trolling for attention. But for those who might think you're serious, what it actually says is that sample-rate has to be at least fast enough to preserve the things we can hear, and even faster if there are ultrasonic noises that can't be removed by filtering. The reason for that is because inadequate sample-rate causes ultrasonic frequencies to fold (alias) back into the audible range, where we do hear them. --Bob K (talk) 15:02, 20 January 2024 (UTC)


 * How does one troll for attention when nothing points to me IRL - including dynamic IP. Audiophool techno-babble is like a weed that needs to be uprooted before it grows bigger. The user above (86.41.241.197) has a real concern that the same techno-babble is growing in this article and yet is not removed because people who think they know, don't, and don't know when to shut up... hence: "...wish Wikipedia was still an encyclopedia worth a damn." Address why wikipedia is becoming irrelevant then you can understand why knowledgable people are frustrated. — Preceding unsigned comment added by 61.69.238.160 (talk) 15:15, 5 March 2024 (UTC)