Fish Piper

I’ve been doing some work at UNSW over the college holidays, getting my brain back into some electrical engineering. Its taken some doing, let me tell you. I have found out that I used to know maths, but now…

One of the very cool things about work has been reading lots of different types of papers in a variety of research fields. I’ve been doing some editing of papers, largely to do with signal processing. This is a field that essentially has to do with taking some sequences of information, then manipulating and tweaking it to acheive a result that you want. Audio processing is one example, coding of image data (such as JPEG) is another. My undergraduate thesis was in this area, where I developed an algorithm that would work with cochlear implants to reduce background noise so that recipients of the device would be able to understand speech better.

So I’ve been able to keep up with some of the things that other people in this research group have been up to, generally related to speech processing. There’s one guy who has done some work in the field of automatic language identification, specifically classification of languages into tonal (languages such as Chinese or Tamil) and non-tonal (e.g. English) categories. He’s pretty much the first person to look into this area, which is exciting. Another guy is developing some speech enhancement techniques that mimic how the human auditory system hears sound, specifically the phenomena that when a loud sound is present, you simply don’t hear softer sounds. Very nifty, and very interesting.

Over the coming week I’m going to have to give myself a crash course in some sophisticated statistical modeling techniques, particularly kalman filtering and particle filters. I’m reviewing a paper that uses these techniques but is a little scant on the detail, and I need to flesh out what is happening a little more. Should be fun, but perhaps a little taxing on the old holiday-brain.