Talk:Linear pulse-code modulation: Difference between revisions

Revision as of 17:15, 25 April 2011

Latest revision as of 00:27, 2 March 2014

(5 intermediate revisions by 2 users not shown)

Line 1:

{{merged-to|Pulse-code modulation|date=2014-03-01}}

== Zero value in signed code==

The article sais ''If the sample is 16-bit signed, the sample range is from -32768 to 32767, with a centerpoint of 0''. If the entire range <math>[-2^{N-1}, 2^{N-1}-1]</math> is allowed, then the zero value actually is (assuming the zero value is the average of maximum and minimum)

<math>

\frac{-2^{N-1} + 2^{N-1}-1}{2} = 2^N\frac{-2^{-1} + 2^{-1}-2^{-N}}{2}=-2^N 2^{-N}/2=-1/2\neq 0

</math>

==Merge with Pulse Code Modulation==

Line 13 ⟶ 8:

___ <span style="font-size: smaller;" class="autosigned">—Preceding [[Wikipedia:Signatures|unsigned]] comment added by [[Special:Contributions/67.122.114.28|67.122.114.28]] ([[User talk:67.122.114.28|talk]]) 03:02, 19 August 2005 (UTC)</span><!-- Template:UnsignedIP2 -->

:I agree -- I'm after a standard, ISO something something -- very much related to computer audio (DLNA specifically) so it's computer-type information that I'm after. <span style="font-size: smaller;" class="autosigned">—Preceding [[Wikipedia:Signatures|unsigned]] comment added by [[Special:Contributions/210.11.153.86|210.11.153.86]] ([[User talk:210.11.153.86|talk]]) 09:16, 31 August 2005 (UTC)</span><!-- Template:UnsignedIP2 -->

:I support a merge. I don't see much support for LPCM being a distinct file format. The references cited use the term "linear audio" but not "linear PCM". These are generic audio formats and the individule files already have their own articles, e.g. [[WAV]], [[AIFF]], [[Au file format]] -—[[user talk:Kvng|Kvng]] 13:47, 27 September 2012 (UTC)

{{done}} ~[[user talk:Kvng|KvnG]] 00:27, 2 March 2014 (UTC)

== Source of the LPCM stub ==

Line 19 ⟶ 18:

The stub for LPCM was taken from an ad for a computer program from the company Cyberlink. This is the website:

http://www.cyberlink.com/english/dv-entertainment/articles/lpcm.jsp

<span style="font-size: smaller;" class="autosigned">—Preceding [[Wikipedia:Signatures|unsigned]] comment added by [[Special:Contributions/65.87.26.127|65.87.26.127]] ([[User talk:65.87.26.127|talk]]) 17:31, 28 September 2006 (UTC)</span><!-- Template:UnsignedIP2 -->

== vs. PCM ==

I think this article should explain the difference between PCM and LPCM.

Line 74 ⟶ 75:

</blockquote>

But this is wrong. The Nyquist Theorem says that the highest frequency of signal that CAN be recorded is 1/2 the sampling rate. It does not say that a sampling rate of R will accurately record a signal of frequency R/2. For example, to accurately record a frequency of, say, 22 kHz with a 44 kHz signal requires that the samples be taken at exactly the maximum of the peak and exactly the minimum of the valleys; if the 22 kHz signal is phase-shifted at all from the period of the sampling, then the 22 kHz will be aliased as a lower-frequency signal by the 44 kHz model. <small><span class="autosigned">—Preceding [[Wikipedia:Signatures|unsigned]] comment added by [[User:MarkRLindsey|MarkRLindsey]] ([[User talk:MarkRLindsey|talk]] • [[Special:Contributions/MarkRLindsey|contribs]]) 01:06, 2 February 2009 (UTC)</span></small><!-- Template:Unsigned --> <!--Autosigned by SineBot-->

== Zero value in signed code==

The article sais ''If the sample is 16-bit signed, the sample range is from -32768 to 32767, with a centerpoint of 0''. If the entire range <math>[-2^{N-1}, 2^{N-1}-1]</math> is allowed, then the zero value actually is (assuming the zero value is the average of maximum and minimum)

<math>

\frac{-2^{N-1} + 2^{N-1}-1}{2} = 2^N\frac{-2^{-1} + 2^{-1}-2^{-N}}{2}=-2^N 2^{-N}/2=-1/2\neq 0

</math>

<span style="font-size: smaller;" class="autosigned">—Preceding [[Wikipedia:Signatures|unsigned]] comment added by [[Special:Contributions/90.229.142.67|90.229.142.67]] ([[User talk:90.229.142.67|talk]]) 14:09, 31 December 2010 (UTC)</span><!-- Template:UnsignedIP2 -->

==Strictly speaking, the term "linear quantization" is self-contradictory==

This article says that "LPCM is PCM with linear quantization". However, there is no such thing as linear quantization. Quantization is an inherently non-linear process. Linear processes are invertible (exceptordinarily) in the degenerate special case)invertible. Quantization is not invertible. No quantizer is linear. Some abuse of basic mathematical concepts is necessary to come up with such a term. This strange term "linear quantization" should be removed, or at least explained. The referenced document does not provide a definition of this self-contradictory term. —[[User:SudoMonas|SudoMonas]] ([[User talk:SudoMonas|talk]]) 16:51, 25 April 2011 (UTC)

:Linear here refers to the step size use for encoding. An example on non-linear PCM is something encoded using [[Mu-law]]. -—[[user talk:Kvng|Kvng]] 13:47, 27 September 2012 (UTC)

== Contributions needing work ==

{{u|Doorknob747}} contributed the following to the article. I'm pulling these contributions here for discussion before inclusion.

I'm not aware of 32-bit linear PCM in consumer applications. 32-bit floating point is a thing. Do you have a citation?

As for the second contribution, discussion of subjective sound quality doesn't usually lead anywhere productive. Especially so for uncited discussion. ~[[user talk:Kvng|KvnG]] 20:30, 21 December 2013 (UTC)

:<nowiki>==Linear 32bit PCM==

There is a L32 bit PCM, and there are many sound cards that support it.

===Similar effects of DTS-HD master audio (192kHz DTS)===

It is said that there is no difference at all that can be heard from a Linear 32 bit PCM at a 96&nbsp;kHz sample rate playback sound to a, high quality DTS-HD Master Audio (192&nbsp;kHz DTS). They both sound the same because, of the extremely high quality sound playback from these two types of codecs. But, in reality the 192&nbsp;kHz DTS sound file has actually 1.45 times better quality than a sound file of L32 at 96&nbsp;kHz.</nowiki>