User talk:Olli Niemitalo

Additive synthesis

Well done with the timeline. I think it looks great. Just heard about Beauchamp's Harmonic Tone Generator from a friend.. you found it before me :) Ross bencina (talk) 01:31, 5 February 2012 (UTC)[reply]

Thanks Ross! It was quite hard to balance. Long time no see by the way, when was that sinusoid fun anyhow -- must be many years ago. Anyways, nice to have you around. :-) (Not saying you should be here day and night like me. :D) Olli Niemitalo (talk) 06:06, 5 February 2012 (UTC)[reply]

Space before <ref>

The best way to keep a special space before the <ref> is to use " <ref>". Another solution in my opinion is use something real before the note number, eg: "...appear in note<ref>..." -- Basilicofresco (msg) 20:45, 18 February 2012 (UTC)[reply]

Sure,   works. Nevertheless consider "note<ref>...", in my opinion it is easier to understand. I suggest you to can ask for an opinion in Help talk:Footnotes.-- Basilicofresco (msg) 07:51, 19 February 2012 (UTC)[reply]

Discussion on the AFT5 Request for Comment

Hey Olli Niemitalo - this is to notify you that there is a discussion starting on the Article Feedback RfC talkpage that has ramifications for the RfC itself. Your input is much appreciated :). Thanks! and apologies if I've missed anyone Okeyes (WMF) (talk) 16:46, 28 January 2013 (UTC)[reply]

Window function

Thanks for all your recent help. Very nice work. --Bob K (talk) 20:47, 11 February 2013 (UTC)[reply]

Glad to hear that from you, I appreciate it. You have done much quality work yourself. I'll see you around! Olli Niemitalo (talk) 22:28, 11 February 2013 (UTC)[reply]

I no longer have access to Matlab, so I'm considering Octave (inspired by your example). But apparently you encountered significant bugs, and that's also the vibe I'm getting from internet chatter. Any advice you have would be appreciated. --Bob K (talk) 17:04, 14 February 2013 (UTC)[reply]

I have not used these programs for very long, only for a week or so, but I'll try to give some advice. Octave is not very stable but if you don't "do mistakes" then it doesn't seem to crash. I use Notepad to edit my commands and then paste into Octave by right-clicking, so I won't lose anything if it crashes.

Installation -- If you are using Windows then I recommend to use the latest MinGW version installer listed here. I had problems with the Visual Studio -compiled version. The installer comes as two .7z files. First install Octave from one of them and after that the packages from the other (the very useful "signal" package is there for instance). Follow the instructions in the readme.

Plotting -- Octave allows you to choose between two plotting libraries. I found that for bitmaps the default one is perhaps better but for vectors (SVG) Gnuplot is better. See the rectangular window image for commands that change the graphics toolkit.

Special characters in figures -- If you want to generate figures with Greek symbols then it seems best to edit the SVG files (they are plaintext XML files so easy to edit) after creating them in Octave and to insert Unicode escape codes for any special characters. There is a TeX interpreter that can be enabled for text labels in figures, but the font that it used was not available at Wikimedia Commons for SVG rendering.

Other programs -- There is also R, which is an environment intended for statistical computing, has a matrix-based language and is more mature than Octave. The signal package is available for R as well. Olli Niemitalo (talk) 21:44, 14 February 2013 (UTC)[reply]

I was successful with Octave and qtOctave (resulting in Overlap-save_algorithm.png). More tips:

The "graphics_toolkit gnuplot" statement also avoids a nasty bug encountered by qtOctave users. You can make the change permanent with this procedure:

Open file (location of your octave directory)/share/octave/site/m/startup/octaverc in a text editor.
Find line graphics_toolkit('fltk') (line 14 in my 3.6.2 Octave distribution).
Change fltk to gnuplot.

I also didn't need the "pkg load signal" statement, I think because I followed these instructions in the 3.6.2 README file:

pkg rebuild -auto
pkg rebuild -noauto ad windows
pkg rebuild -noauto nan % shadows many statistics functions
pkg rebuild -noauto gsl % shadows some core functions
pkg rebuild -auto java

Thanks for your help and encouragement.

You're welcome! I did something like that too, but left "pkg load signal" in the script for others who might want to run it. Olli Niemitalo (talk) 18:02, 17 February 2013 (UTC)[reply]

--Bob K (talk) 15:40, 17 February 2013 (UTC)[reply]

Window figures

Good work on updating the figures for window functions. I've just added details of the Planck-Bessel window, and I was hoping you might be willing to add a figure for that. I don't have an installation of Octave. I can confirm your code doesn't work with Matlab, and I thought there isn't much point tweaking it as the results won't be quite the same. I think the snippet of code you want is:

tau = N-1;
alpha = 14;
epsilon = 0.1;
t_cut = tau * (0.5 - epsilon);
t_in = k - 0.5 * tau;
T_in = abs(t_in);
z_exp = ((t_cut - 0.5 * tau) ./ (T_in - t_cut) + (t_cut - 0.5 * tau) ./ (T_in - 0.5 * tau));
sigma =  (T_in < 0.5 * tau) ./ (exp(z_exp) + 1);
w = (1 * (T_in <= t_cut) + sigma .* (T_in > t_cut)) .* besseli(0, alpha * sqrt(1 - (2 * t_in / tau).^2)) / besseli(0, alpha);
plotWindow(w, "Planck&#8211;Bessel", "Planck-Bessel", "&#949; = 0.1, #&945; = 15", "epsilon = 0.1, alpha = 14")

— BobQQ (talk) 18:27, 19 February 2013 (UTC)[reply]

File:Window function and frequency response - Planck-Bessel (epsilon = 0.1, alpha = 14).svg

Planck–Bessel window, α = 14, ε = 0.1; B = 2.16.

Thanks! No problem, I made the image and can do others/modify if needed. You can add it to the article from here if looks OK. The B value is what Octave reported. I corrected a couple of certain typos (alpha value in image title: 15 --> 14, fix swapped # and &) and put the source code in the Rectangular window image description. Olli Niemitalo (talk) 19:48, 19 February 2013 (UTC)[reply]

Oops, I think the Kaiser function calculation should have pi * alpha instead of alpha, to be consistent with the rest of the Window function article. Olli Niemitalo (talk) 20:03, 19 February 2013 (UTC)[reply]

I had a look at the Berry & Gair 2012 gravitational wave burst article where the window was discussed. They have beta = 14, which agrees with the definitions of alpha and beta, by JOS. So I suggest a few alternative approaches to resolve this:

1) Create a new image with the parameter name changed to beta (Kaiser window section must then explain the difference between alpha and beta)

2) Create a new image with alpha = 3 (used as beta = pi*alpha = 9.42477796077 in besseli) for easy comparison with Kaiser window

3) Create a new image with alpha = 4.5 (which gives beta = 14.1371669412) so that we are quite surely in the use region of the astronomy paper

I'll make a figure along alternative 2 ~~(my favorite)~~ while waiting for your comment. Olli Niemitalo (talk) 20:35, 19 February 2013 (UTC)[reply]

File:Window function and frequency response - Planck-Bessel (epsilon = 0.1, alpha = 3).svg

Planck–Bessel window, α = 3, ε = 0.1; B = 1.80.

Oops. I added irrelevant info here about Kaiser window. I was confused. And now I've removed it. Sorry! --Bob K (talk) 12:52, 20 February 2013 (UTC)[reply]

I played with this a bit and it seems that pi*alpha = 14 is sort of a lucky value for which the first side-lobe lines up nicely with the rest of the side-lobes. I don't think it's exact in any sense. The value alpha = 4.5 doesn't give quite as pleasing a result (looks somewhat like alpha = 3 but not nearly as severe), but alpha = 4.45 looks good. I'll make an image with that value even though the value is a bit cumbersome.

Going on a tangent, there is an exact formula for the bandwidth of the Kaiser main lobe in Kaiser and Schafer 1980 - On the Use of the I0-Sinh Window for Spectrum Analysis. It can be rewritten as alpha = sqrt(c^2 - 1), where c is the location of the first null (in bins). alpha approaches c asymptotically at large values of c. But that doesn't seem to help us here as Planck-taper messes up the situation. Olli Niemitalo (talk) 08:06, 20 February 2013 (UTC)[reply]

Thanks, that's a prompt response! Good work on spotting the factor of pi, I obviously hadn't correctly noted the differences in convention between the paper and the rest of the article. I'd say using the value of 4.45 so πα = 14(ish) is the best option. If you think that there is some interest behaviour when varying α or ε, perhaps it would be worth illustrating with a second figure? Might take up more space than warranted, but we do have two figures for the Kaiser and exponential windows, and they only have one parameter. — BobQQ (talk) 09:50, 20 February 2013 (UTC) (PS Thanks for the shared credit, it is appreciated, but unnecessary since I only produced a couple of typo-filled lines... The perils of writing code you can't test)[reply]

Let's see if a source that uses different values turns up, I don't want to go improvising. Olli Niemitalo (talk) 14:10, 20 February 2013 (UTC)[reply]

Star

		The Graphic Designer Barnstar
		For fixing this, and for your many other image contributions. *Arbitrarily0* ^(talk) 14:33, 11 March 2013 (UTC)[reply]

Orphaned non-free media (File:Synclavier II Demo 2.ogg)

Thanks for uploading File:Synclavier II Demo 2.ogg. The media description page currently specifies that it is non-free and may only be used on Wikipedia under a claim of fair use. However, it is currently orphaned, meaning that it is not used in any articles on Wikipedia. If the media was previously in an article, please go to the article and see why it was removed. You may add it back if you think that will be useful. However, please note that media for which a replacement could be created are not acceptable for use on Wikipedia (see our policy for non-free media).

If you have uploaded other unlicensed media, please check whether they're used in any articles or not. You can find a list of 'file' pages you have edited by clicking on the "my contributions" link (it is located at the very top of any Wikipedia page when you are logged in), and then selecting "File" from the dropdown box. Note that all non-free media not used in any articles will be deleted after seven days, as described on criteria for speedy deletion. Thank you. Hazard-Bot (talk) 04:22, 24 May 2013 (UTC)[reply]

A free alternative was found. Olli Niemitalo (talk) 05:43, 29 August 2013 (UTC)[reply]

Article Feedback Tool update

Hey Olli Niemitalo. I'm contacting you because you're involved in the Article Feedback Tool in some way, either as a previous newsletter recipient or as an active user of the system. As you might have heard, a user recently anonymously disabled the feedback tool on 2,000 pages. We were unable to track or prevent this due to the lack of logging feature in AFT5. We're deeply sorry for this, as we know that quite a few users found the software very useful, and were using it on their articles.

We've now re-released the software, with the addition of a logging feature and restrictions on the ability to disable. Obviously, we're not going to automatically re-enable it on each article—we don't want to create a situation where it was enabled by users who have now moved on, and feedback would sit there unattended—but if you're interested in enabling it for your articles, it's pretty simple to do. Just go to the article you want to enable it on, click the "request feedback" link in the toolbox in the sidebar, and AFT5 will be enabled for that article.

Again, we're very sorry about this issue; hopefully it'll be smooth sailing after this :). If you have any questions, just drop them at the talkpage. Thanks! Okeyes (WMF) 21:54, 1 September 2013 (UTC)[reply]

A barnstar for you!

	The Original Barnstar
	Thank you for making and uploading all those different window function graphics. I found them very helpful. ~Adjwilley (talk) 22:34, 11 July 2014 (UTC)[reply]

Triangular window

"Window function and frequency response - Triangular.svg" really replaces "Window_function_(triangular).png"? See: this page

Daroooo (talk) 22:12, 9 July 2015 (UTC)[reply]

The SVG image was created to replace the PNG. In the process of converting the window function images, the bin axis was zoomed in a bit to better show the main lobe and the first sidelobes of the window functions. The frequency response axis was later renamed Fourier transform, which is more correct because the window function is not a system. Otherwise the two images are interchangeable. It seems that not all wikis have yet migrated to using the SVG window function images. Was there something else? Olli Niemitalo (talk) 07:14, 13 July 2015 (UTC)[reply]