Jump to content

Talk:Hann function

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Name

[edit]

Isn't it a little overly opinionated to refer to the term "Hanning window" as "erroneous" and "confusion"? It is in widespread use and has become a (arguably the) standard term for the function. The Matlab function to generate one is called hanning(), for example. I was well aware that the man's name is Hann, not Hanning, but I had always interpreted the name of the window as more of a joke or play on "Hamming" than a genuine misunderstanding, along the same lines of how the term "nybble" was coined as a play on "byte". Even MathWorld, which is linked as a reference, calls it Hanning. Anyone have a reference to back up the claim that it's a real mistake? 146.6.203.192 (talk) 00:21, 3 March 2010 (UTC)[reply]

Merge into main article Window function

[edit]

From the main article only two types of windowing functions have their own page, the Hann function (this page) and the Kaiser window. Of the two this page does not provide any further information than in the main article.

Furthermore it it in its current form it is incorrect. The formula is clearly wrong (clearly S must be a function of τ) and none of the variables are defined (what are S? f?).

Should this page be removed so the Hann function is only defined in the main article or expanded in depth to be useful?

Nonnumquam (talk) 07:31, 30 November 2009 (UTC)[reply]


Incorrect formula

[edit]

First time writer, so just talking rather than editing.

The formula given is wrong - better to use 0.5*[1-cos(2pi*[n+0.5]/N)]

Just compare for, e.g., N=1 (a perfectly valid choice for which the only value of n is 0).

Misleading statement

[edit]

I think this is misleading:

"If the Hann window is used to sample a signal in order to convert to frequency domain, it is complex to reconvert to time domain without adding distortions."

Isn't that a general property of the short-time Fourier transform, regardless of what window you use?

Cryptochromatologist (talk) 18:37, 16 November 2009 (UTC)[reply]

I agree. It is unclear what this comment even means; reconstructing a signal from a truncated version is impossible without some assumptions about the signal, it is not a property of the Fourier transform or windowing function. Nonnumquam (talk) 07:31, 30 November 2009 (UTC)[reply]


Another agree. In fact if you use windows with 50% overlap the hann window is convenient to get back to the time domain by simply adding the ifft of fft of the overlapping chunks. —Preceding unsigned comment added by 77.102.249.198 (talk) 22:47, 19 April 2011 (UTC)[reply]

Channel numbering

[edit]

The formulas seem to refer to a situation with channels 0,1,2,...N-1 where the outer channels are given zero weight. If only non-zero channels 1,2,....N are considered the formulas will be sin2(pi*n/(N+1))=1/2*)1-cos(2*pi*n/(N+1)).150.227.15.253 (talk) 14:59, 21 February 2019 (UTC)[reply]

Please define constant variable L

[edit]

I'm a math major, not an engineer. I must assume "L" in the formula refers to sampling frequency, but it would be nice if this were explicitly stated. — Preceding unsigned comment added by 65.130.87.104 (talk) 00:12, 27 October 2019 (UTC)[reply]