In light of the ongoing disruption that Coronavirus poses to our lives, which includes the cancellation of our annual MST Research Symposium, we are publishing a series of posts on MST students and their research that have been peer selected for excellence.
We asked each group selected to give us info on their abstract, the biggest takeaway from their project study, and something they are excited about for the upcoming year, as well as a few photos. This post covers the work done by MST sophomores Ashton Doane, Chris Spencer, and Sunny Wang.
Abstract: Sophomores Ashton Doane, Chris Spencer, and Sunny Wang conducted research on the applicability of neural networks to music transcription. Music transcription is the task of taking some audio and creating sheet music, a notation of how to play it on an instrument. The overall goal of the project was, “to achieve great accuracy in transcribing polyphonic music through machine learning, hopefully having a higher maximum potential for transcription than humans.” Using some key insights into the inner workings of musical structure, they created multiple models to attempt to increase accuracy. Their results can be found here: Automating Music Transcription with Artificial Intelligence.
When asked about their biggest takeaways from their project study this year, they said it “… was that a lot of preparation work must be done to effectively finish a task, yet there still needs to be room for adjustment when things go off course. Making sure that you can make a plan and stick to it without cheating or taking days off makes sure that you can finish a project.”
As for next year, they said, “Internships are approaching, which is very exciting, as they give an opportunity to work in a real scientific environment. The real excitement here is in getting new ideas and seeing if a certain job is a good fit for people. The opportunity to both try out a job and meet new people is a great advancement in life.”