Wednesday, May 31, 2006

June 1

Now it comes to the beginning of another month. This means I only have one month left and I am deperately in need of RESULTS!!! Gimme some RESULTS!!! argh!!!

I've been struggling fiercely with visAD and feel like wading through a marsh. I can't figure out what the problem is. I've been trying different setting of classpaths, moving the folder from here to there, but to no avail. I think I should give me the idea of using visAD to display graphs.

No more slacking please.

--

Later on I decided I should stop mulling over the visAD stuff and turned into refining what I have done. A couple of preliminary results show something awkward, perhaps I'm on the verge of jeopardizing my whole idea about Tristimulus. I hope training with HMM can help conceive some more promising results. Thanks God I found a tool called Jahmm and as soon as I can get through the problem with the format with the input sequence I am DONE! (well... hopefully)

Monday, May 29, 2006

May 29

I broke my promise. I didn't report my progress daily over the past few days. Possibly because my progress is so slow that I am ashamed to tell you how it's going.
I can't always dodge my duty (by dozing off, or keep checking xanga blogs, for example). Here I write down what I have done in the past three working days. To sum up, I have transformed the matlab code for tristimulus to a Java code. I spent (more precisely, wasted) days to look for codes for FFT and those math stuff. Finally I have looked for some work-around. The problem is that the code for FFT that I have adopted on work for window lengths of powers of 2. Consequently I can't compare the results directly without modification. I'll leave the work for tomorrow. I also look forward to finding a more sound way to look for harmonics. Now it's jogging time. See ya.

Tuesday, May 23, 2006

Directions suggested by Edward

1. Train the existing algorithm with a larger set of songs, confined to a narrower genre (Canto-pop)
2. Users' perspective - selection of songs by affinity/type

May 23

It's the first day of the FYP extension. The progress simply staggered. Should I attribute to my lack of sleep last night or unfamiliarity of the JAVA libraries? Anyway I have only completed some very simple math functions including mean, variance and hanning function (easy-as-pie jobs). I have identified a couple of tasks to be completed within this week. First of all I have to learn stuff to manipulate wave files. Also I have to look for an algorithm to find harmonics more precisely. I hope I can learn something about HMM and the ultimate goal is to apply HMM in my project.

Monday, May 22, 2006

The Creation

In the beginning God created the heaven and the earth. And the earth was without form, and void; and darkness was upon the face of the deep. And the Spirit of God moved upon the face of the waters. And God said, Let there be light and there was light.

God's creation is wonderful. As a programmer I am doing something alike (how dare I say that!). Well, though my power is not a bit comparable to the almighty and transcendental creator I'll try my best to straighten out everything and make my little oeuvre work.

I'll try to be honest and jot down my progress in the blog every day. Hold on, don't let my supervisors ever step onto this place =P