Hi I’m Yifan, a senior studying Computer Science at University of Michigan. I’m working on the Rainforest Wildlife Recording Project. This project is a continuation of 2017 fellow Zach’s Songbird program. Me and my teammates also worked on this during last semester, and will continue to improve it in the coming semester. We have already made a prototype device with wifi connectivity, but our hardware engineer brought it with him back to China. Thus for now, I’ll use last year’s device to get bird recordings in the wild.
This summer I am aiming to mainly work on expanding the functionality of our identification algorithm. The previous device uses a threshold trigger recording strategy, which means the device only starts recording when it detects a sound. However, this mechanism will sometimes miss valuable sounds; since we don’t want to miss any awesome bird songs, we’ve made the decision to switch to continuous recording. Continuous recording will produce a huge recording on a daily basis. One of the biggest challenge for me would be processing this large chunk of recording, and present (even classify) the interesting data to our user.
For this week I will be working on getting the old device to work (it’s not functional right now). After that I will be using it to get some bird recordings in the wild (yay). Luckily, Ann Arbor in summer has a large songbird population, so hopefully that will go smoothly. In the meantime, I can do more research on the classifying and interest-detecting algorithm. Furthermore, I can also see myself working on building a new case for our device in the future, because the one we have now is hardly waterproof (for now).
Hopefully by the next blog, I can show some actual data recorded from the device and processed by the software. Wish me luck!