MakeHMMPrototype, written by David Dean, 2004

Generates a forward only, diagonal covariance HMM prototype

usage:
   MakeHMMPrototype name parameterkind vectorsize numberstates mixturesperstate \
                [ streamcount vecsize1 ... vecsizeN ]

   name             : name of hmm (eg. prototype)
   parameterkind    : parameter kind (eg. MFCC_0_D_A)
   vectorsize       : size of feature vector
   numberstates     : total number of states
   mixturesperstate : number of mixtures per state
   streamcount ...  : multiple-stream support
                      (vecsize1 + ... + vecsizeN = vectorsize)

You can download the c code to compile to your favorite platform here.

It was really simple to do:

1. Point davidbdean.com at this website
2. Create a redirection website with only a .htaccess file containing ‘RedirectPermanent / http://www.davidbdean.com/‘
3. Point cebidae.com and dbdean.com to the redirection website

And it all just, hopefully, works.

But, in the meantime, I thought I’d point out how I go about actually reading ebooks rather than just aquiring them. I’ll start with the reading software, and work back to ebook formats.

μBook

If you want to read ebooks comfortably then (in my opinion) μBook the best choice, and the only software you should need. It has support for the most formats, runs on Windows, Pocket PC and now has a ‘lite’ Palm version (Sorry no *nix/OSX – any alternative suggestions in the comments?). It’s not free, but at $15 US I think it is worth it.

In my experience, the main differentiating factor with ebook software is support for formats. Microsoft Reader and eReader only appear to support their own, proprietary formats, which are normally DRMed. Does this seem silly to you? An ebook reader prorgam that cannot even open a text file for reading? Crazy.

Mobipocket has support for a few more formats, such as RTF, HTML and TXT, but μBook has support for many more:

• Renders .TXT, .RTF, .HTML, .PML, .PDB and .PRC (Non-Secure), files book-like.
• Extracts text out of .PDF (Non-Secure) files.
• Displays PRC, RTF and HTML images, .JPG, .GIF, .PNG, .WMF (in RTF) or .BMP.
• Opens text inside .ZIP, .CHM and .RB files, uncompressing paragraphs on the fly, limiting the amount of memory required (except CHM).

It’s that last point that seems most important to me. I think ebooks should be in the most open format that can support all the features required. To me, that would probably be HTML, but with multiple pages, images and style sheets you can end up with a few files per ebook. Μbook (Μ is a capital μ) allows you to zip them up so they can be in a single file.

So, anyway, on to ebook formats.

Ebook Formats

First off the bat, Adobe PDF (as an ebook format) is evil. Very evil. No good unless you want to read them on a screen that is at least as big a sheet of paper. I normally read ebooks on my Pocket PC or my Tablet PC so I need my ebooks to be in a format that can be read on any size screen, and PDF does not cut it in this regard. I normally use Adobe Acrobat to convert the PDFs to something more useful, normally RTF or text. If the PDF is locked so I cannot save it as something else then I don’t read it – too much trouble.

If I’m buying an ebook from somewhere online, like Fictionwise, I normally get (unsecure, DRM-free) Palm Doc (pdb) or Mobipocket (prc) versions, as they work well in uBook.

If I’m getting some short fiction from the web somewhere, I normally try and find a ‘printer-friendly’ format and save the HTML file. If the story has important images, then I use Firefox’s ’save as -> web page, complete’ option to save the page to a folder, and zip that up.

One final suggestion if a website has too much junk around the fiction: Use the Scrapbook extension for firefox. It allows you to save a selection to a file, including all the images.

To the future: OpenReader maybe? Sounds good, but so far its been mostly vaporware its been a long time coming – David’s comment below has details. I really think there needs to be a free and open standard in this space, and I hope it’s not too far away.

HIST2D calculates a 2-dimensional histogram

N = HIST2D(X,Y) bins the X and Y data into 10 equally spaced bins in both dimensions

N = HIST2D(X,Y,M), where M is a scalar, bins the data into M equally spaced bins in both dimensions

N = HIST2D(X,Y,B), where B is a vector, bins the data with centers specified by B

The number of bins or centers can be specified individually for either dimension using N = HIST2D(X,Y,NX,NY) or N = HIST2D(X,Y,BX,BY)

[N,BX,BY] = HIST2D(…) also returns the positions of the bin centers as two matrices in BX and BY

HIST2D(…) without output arguments produces a colormapped image plot of the 2d histogram

EXAMPLE

yin = randn(1,1000);
xin = randn(1,1000);
[n,x,y] = hist2d(xin,yin,11);
imagesc(x(1,:),y(:,1),n);
hold on;
plot(xin,yin,'y.'); colorbar

You can download it here: hist2d.m

1. Make sure the harvard package is installed in Latex.
2. Copy the CRPITStyle.cls class file available here to your localtexmf/tex/latex, and make sure you refresh the Latex database
3. Copy CRPITStyle.layout into Lyx’s layout folder (C:\Program Files\LyX141\Resources\layouts for me)
4. Open Lyx, Run Tools -> Reconfigure (you may also need to use Tools -> Tex Information -> Rescan)
5. Unzip this file somewhere
6. Open CRPITPaper.lyx
7. Modify the file to your hearts content (this one is setup for VisHCI 2006)

Let me know if this doesn’t work for you, as it took me some mucking with refreshing Latex and Lyx to get it to work, so I may have missed something in my re-telling.

Word of warning tho: It says

Web-based install: In WaveSurfer, select Plug-ins from the File menu. Under Plug-ins available for installation, select video and click install.

I don’t know where that is. I have the latest version of Wavesurfer (I just downloaded it) and no such option exists in the file menu. Just follow the manual install instructrions (which I have edited slightly – in bold)

Manual installation (Windows): Download and open http://www.speech.kth.se/wavesurfer/video.zip . Put the files video.plug and QuickTimeTcl.dll the WaveSurfer plugins directory %HOMEDRIVE%%HOMEPATH%\.wavesurfer\1.8\plugins for WaveSurfer 1.8.*, where HOMEDRIVE and HOMEPATH are Windows environment variables (in most cases the directory will be C:\Documents and Settings\username\.wavesurfer\1.8\plugins). Optionally, also put Video.conf in your configurations directory.

UPDATE: I guess this is an example of freedback, so I’ve added that category/tag. (More freedback here)

First Create a new layout file called spconf.layout in your Lyx layouts folder (in my case “C:\Program Files\LyX141\Resources\layouts”) with the following contents:

#% Do not delete the line below; configure depends on this
#  \DeclareLaTeXClass[article,spconf.sty]{article (spconf)}

Input article.layout

Preamble
        \usepackage{spconf}
EndPreamble

This will create a new document layout in lyx called “article (spconf)” that will be identical to article, but adds \usepackage{spconf} in the document preamble.

Now we need to fix up some of the header information. spconf uses \name instead of \author, so we’ll add that to our file:

Style Author
        LatexName       name
End

And finally, we need to add an affiliation section that uses \address in LaTex, and is used when the \maketitle command is used:

Style Affiliation
        LatexType       Command
        LatexName       address
        InTitle         1
End

There, that should work. Make sure that spconf.sty is stored where latex can find it (normally the same folder as your lyx file). If you want to download the entire spconf.layout file it is available here.

In case you haven’t noticed this blog used to be at cebidae.blogspot.com, but I decided to move it to my own server so I had more control over what I could do, now, and in the future.

Importing the old blog’s entries was a trivial task as it is build into the wordpress software these days. The main difficulty I was having is how to get people arriving at the old site to come over here.

The main advice appears to be that you should leave your blogspot site there, but redesign so it is clear that you have moved to a new site. I wasn’t really happy about this, because people are lazy, so I wanted to automatically redirect them to the new site.

There are basically three types of pages you want to redirect from your blogspot address to wherever it is you are going:

1. The main page (/)
2. Archive pages (/2004_12_01_cebidae_archive.html)
3. Item pages
   (/2004/10/gadget-price-searching-use-staticice.html)

I wanted to redirect these three types of pages to the same pages on this site. I managed to get the main page and the item pages working but not the archive pages. Seeing as how most external links either point to the main page or a particular item page, the archive pages are unlikely to be landed on too often, so I chose to just redirect them to the new main page here. If anyone can help me redirect the archive pages correctly, that’d be great.

Redirecting the main page (and the archive page)

To set up a redirect to “http://www.cebidae.com” you need to use the following tags between the <head> and </head> in your blogger template

<MainOrArchivePage>
<META http-equiv="refresh"
content="3;URL=http://www.cebidae.com">
</MainOrArchivePage>

MainOrArchivePage makes sure this only happens on the main or archive pages and the meta tag tells whoever lands here to redirect to http://www.cebidae.com in 3 seconds.

Redirecting the item pages

Obviously if you are going to redirect your visitors to a new site, you need to know the url that they should be going to. This was easy with the main pages, but a bit harder with the item pages. I decided that the best option was to use the built in search on this site to search for entries matching the title of the old entry. To do this you need to insert the following tags after </MainOrArchivePage> but still within the <head> and </head> tags:

<ItemPage>
<Blogger>
<META http-equiv="refresh"
content="3;URL=http://www.cebidae.com/index.php?s=<$BlogItemTitle$>">
</Blogger>
</ItemPage>

ItemPage makes sure this only happens on item pages, and this time the redirection is to a search on the new site for the title of the item on the old. the Blogger tags are required or blogger won’t evaluate <$BlogItemTitle$>.

I also put a small body section in my template that has a link to the new site in case their browser doesn’t support redirection:

<body>
<br/><br/>
<MainOrArchivePage>
<p>This page has moved. You can find the new page
<a href="http://www.cebidae.com" >here</a>.
You should be redirected in 3 seconds.</p>
</MainOrArchivePage>
<Blogger>
<ItemPage>
<p>This page has moved. You can find the new page
<a href="http://www.cebidae.com/index.php?s=<$BlogItemTitle$>" >here</a>.
You should be redirected in 3 seconds.</p>
</ItemPage>
</Blogger>
</body>