Katie Wolf - DREU Experience |
- Distributed Research Experience for Undergraduates - Summer 2009 - | ||||||||||||||||||||||
|
Week 1 May 26th - May 31st: The Beginning Week 0 < ----------------- > Week 2
Since it was Memorial Day on Monday, my first day of work did not begin until Tuesday.
I spent Monday chilling out and unpacking my stuff at the house and I walked around
campus to check things out. OSU is so beautiful with flowers everywhere and the cool
old buildings.
Tuesday I got introduced to the team at their weekly meeting and found out the different possibilities
of the project that I may get to work on. The project as a whole focuses on Signal Processing and
Machine Learning. Two things that I really had no knowledge of before this project. So most of
the week was just reading some of the papers that the team has put together on their project and papers
that others had written on the topics. Forrest (the Phd Student on the project) introduced me to the code
he has written so far. Towards the end of the week he gave me a task to do some coding reusing some of
his code to implement a program that sums up the energy in a Spectrogram and outputs the top X% of chunks
of the audio and chunks of the spectrogram that have the greater energy. This was more a project to get me
familiar with working with the code, but it is also one method that we are using to pick out the possible bird
songs in a long recording.
There are two parts of the project that they may have me work on. One is working more on the signal processing
side and coming up with a good method of cleaning up the recordings from the HJ Andrews Research Forest. A lot of
the audio recordings are muffled by streams that lie in the valleys of almost all the hills in the forest. Each of
the recordings from the forest are around 20 minutes long and we want an efficient way to clear up the sound by getting
rid of the background noise so that we can clearly pick out the birds songs and effectively label them to the species of
bird. The other part of the project that I may get to work on is creating an interface for the ecologists to work with
that allows them to manually classify audio clips to add to the training set of recordings. I have never worked with
any GUI applications or any sort of signal processing so the whole project seems kinda of overwhelming.
WEEKEND: Living off of campus has been so nice, and I am glad I ended up in the situation I am in. If I were to live in
the dorms, I don't think it would be bad, especially right now, but the semester will be ending soon and I guess most of the
students end up leaving the town. But my roommates will still be here, and we will continue to have a good time, hopefully.
On Saturday I went with Forrest and two of the ecologists on our team to visit the research forest where we are collecting our
data from. I got to see how the devices they are using to collect the audio data and the environment in which the recordings
are coming from. It was a really cool experience. Pictures of the song meters are found in the photo gallery.
|