Automatic music transcription

• المؤلف الرئيسي: Yang, Qihuang
• مؤلفون آخرون: Lin Weisi
• التنسيق: Final Year Project
• اللغة: English
• منشور في: 2016
• الموضوعات: DRNTU::Engineering
• الوصول للمادة أونلاين: http://hdl.handle.net/10356/66736

In the recent years, many methods on automatic music transcription are proposed. Due to interests in Harmonic Adaptive Latent Component Analysis (HALCA) model and its application to music transcription, in this project report, I reviewed the works about the HALCA model theory and related previous works in order to find any possible ways to improve the performance of the music transcription program based on HALCA model. My work can be divided into three parts. Firstly, a new sliced CQT is used to replace the current CQT method in the given program. Then I worked on post-processing algorithm to improve the performance of HALCA to music transcription. Thirdly, an effort is made to find an automatic threshold determination method for extracting information. Finally, limitations of current methods are analyzed and future research directions are identified.