Another area that interests me is in the comparison of a computer's and a human's ability to recognize different types of information. In the case of sounds produced by animals, there are many situations where humans are superior at recognizing these sounds, but this often requires a great deal of experience. I wanted to see if there was some way to develop this ability in a person without taking years of practice. I chose to test this with bird songs because people can hear them, unlike the echolocation calls of bats. It's also well-known that people can get to be quite good at telling one bird from another. I wrote a program in Visual Basic that plays a variety of different bird songs as a training exercise so that the user can learn them. For each sound, the program can display a picture of the bird along with a visual display of the sound called a sound spectrogram. I chose 10 different species of bird that are fairly common in our area so that it would help the user learn to recognize species that they would encounter in this part of Georgia.
The second part of the program then tests the user with some of the same sounds (but with some that the user has never heard before as a control). All the results of the tests are saved in a spreadsheet to make it easy to see how the user is doing, along with which sounds are easier and more difficult to learn. During the training, the user is presented with a sound spectrogram, but no picture of the animal, so the user has to be able to recognize it based on the sounds.
This project was a research project that I began with a couple of different students, but due to a variety of reasons it was never progressed very far, so there are opportunities to actually work with the project as well as updates/improvements to the software itself.