Extracting performance data from great piano performances : filtering and formatting of data

• Main Author: Teo, Qing Xi
• Other Authors: Ong Keng Sian, Vincent
• Format: Final Year Project
• Language: English
• Published: 2016
• Subjects: DRNTU::Engineering
• Online Access: http://hdl.handle.net/10356/67712
• Institution: Nanyang Technological University

As piano recordings on analog form degrade, they are being converted to digital format, one of them being the MIDI format. Such conversion requires an analysis on the recordings that can extract the information needed for a MIDI file, pitch or note being one of the information. Wavelet Transform has been under research to provide a more efficient alternative to the Short Time Fourier Transform for music analysis. This project explores the use of Wavelet Transform to extract the frequency information from a WAV audio file for analysis. The analysis involves pitch detection which detects the notes that are actually played and filters out the harmonics. Although the program written to perform the analysis is able to detect most played notes with reasonable accuracy, it can be further tuned and optimised to obtain better accuracy. While the analysis algorithm can be improved, the Wavelet Transform shows its potential as an efficient alternative for music analysis.