It seems that after reading my previous post, Jorge Cham has finally decided to admit the truth!

I’ve just heard from HCSNet that ten travel bursaries of $500 towards travel and accommodation costs are available to PhD students from outside metropolitan Sydney who wish to attend the NICTA-HCSNet Multimodal User Interaction Workshop, to be held at the Australian Technology Park, Redfern, Sydney, on September 13-14th, 2005.

The workshop includes two invited talks from internationally-recognised researchers in multimodality: Professor Sharon Oviatt from the Oregon Health and Science University, and Professor Francis Quek from the Virginia Polytechnic Institute and State University.

In the interests of information sharing, it is a condition of receipt of a travel bursary that the student should provide a poster describing their current research project.

The closing date for applications for bursaries is Friday August 26th 2005. Those interested should send an email to the HCSNet Convenor (Professor Robert Dale) at rdale at ics.mq.edu.au.

Made sure we had another short paddle on the Lane Cove River this morning. We intended to paddle as far as we could upstream; unfortunately, this was trivially easy, since we couldn’t get any further than the Steakhouse at Fullers Bridge due to the Parramatta Epping to Chatswood Rail Link tunnel works, which have blocked the river. Bugger.

So we ended up paddling up and back between Fullers Bridge and Magdala Park. We both felt pretty sluggish this morning, so we only paddled for about an hour. It rained for much of the paddle, but that didn’t make it any less enjoyable. In fact the sound of the rain on the river was quite soothing.

Have been trying to focus on writing my dynamic time warping speech/speaker recognizer in R this afternoon. Hrm… it’s turning out to be trickier than I thought, but that’s largely because I don’t know R as well as I should. I know exactly what I want to do conceptually – it’s just a matter of working out a smart way to do it in R (like finding out how kmeans clustering works, to avoid having to reinvent the wheel). Back to the grind-stone now. Wish me luck!

So it’s far from finished, but I’ve finally bitten the bullet and published my new site to the web. It’s been sitting on my staging/development server for more than two years now, although the current incarnation bears very little resemblance to that old site!

I’ve focussed on two sections at the moment:
- Trying to create a collective resource that documents how people are using the Enron Email Corpus. This is a massive collection of real-world email from Enron that is available for research purposes. (If you’re interested, head on over to the Enron Email Corpus pages)
- Documenting relevent resources for my Masters Project (which will hopefully lead into a PhD), looking at discourse structures and intention in email communication. This is based around email classification (at least partially), and will hopefully make use of the Enron corpus, both for investigating patterns of communication, and for ensuring that the tools produced work on real-world, noisy data.

Still got a mountain of uni work to do (about 15,000 words of essays for Speech Recognition, as well as constructing a speaker/speech recognition system in R; and a formal literature review and research proposal for my research project). One day I’ll feel like I’m actually making progress!

Finally managed to apply myself well this afternoon and this evening to finish off the gaussian classification assignment for my Speech Recognition Masters subject. Basically, the assignment involved using R and Emu to analyse some speech data (860 short Australian English vowels from 10 different speakers) to train, optimise and evaluate a Gaussian language model using different feature sets. The various feature sets investigated for their utility in classificationwere vowel formants, cepstral coefficients and Bark scaled spectral bands.

Interestingly, although using the first and second vowel formant is a very simple approach, for the particular constrained problem I worked on, they provided the highest accuracy (about 82%). Note, however, that this was only for classifying short Australian English vowels (no diphthongs or long vowels, let alone consonants).

Anyway, it’s nice to have another assignment behind me. The reason I managed to finish it today is because I’m taking a few days off work, in lieu of the massive amount of overtime I put in to get SciFly finished. So hopefully I’ll make some more progress on my other uni work before the week is out!