Unsupervised Transcription Of Piano Music

1835 Words Mar 14th, 2015 8 Pages
Unsupervised Transcription of Piano Music

MS Technical Paper

Fei Xiang
Mar.14, 2015

1. Motivation
Audio signal processing has been a very active research area. Automatic piano music transcription, of all the tasks in this area, is an especially interesting and challenging one. There are many examples of how this technique can contribute to our life. For instance, in today’s music lessons and tests, we often rely on people’s hearing ability to judge whether a piano player performed well based on whether the notes played are accurate or not. The process requires man-power and is not always fair and accurate because people’s judgement is subjective. If a good automatic transcription system can be designed and implemented with high accuracy, teachers can use it to automatically judge a student’s performance assignments.

The task itself is very difficult for a variety of reasons. First of all, piano notes do not consist of a single frequency of sound wave. They have a very rich composition containing a full spectrum of harmonics that have different powers and frequencies. Additionally, the signal properties of piano notes vary from piano to piano. As a result, the traits of piano notes must be learned by computer in an instrument-specific way. Secondly, music, especially piano music, is polyphonic, which means at any given time, its highly likely that multiple notes are played at the same time. This means the harmonics of different notes may collide with each other,…
Open Document