Please use this identifier to cite or link to this item:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/67155
Title: | 影像處理應用於鋼琴琴鍵打譜 Apply Image Recognition to the Musical Composition Using Piano |
Authors: | Yen-Hsiang Chen 陳延祥 |
Advisor: | 丁肇隆 |
Co-Advisor: | 張瑞益 |
Keyword: | 按鍵判別,樂器數位介面,影像處理,訊號分析, Note on detection,musical interface,image processing,signal analysis, |
Publication Year : | 2017 |
Degree: | 碩士 |
Abstract: | 本研究旨在建立一種新的琴譜紀錄方式。以鋼琴為實驗平台,配合當代電腦影像處理的技術,將傳統撰寫鋼琴琴譜的方式從硬體移植到軟體上。本研究的目的是在於提供打譜軟體使用者及音樂教育者能以更低成本,使用規範上更簡易的方式進行樂理中音調及拍數的紀錄。由於錄影設備及電腦使用的普及,當代的樂器演奏往往被以音訊或影像的方式記錄下來,然而除了以音訊的方式,若配合影像處理及電腦演算,其亦可將此種紀錄轉換成樂器學習者所使用的琴譜來呈現。本研究使用網路上以及一些錄製的影片來進行轉譜實驗,配合簡易的影像及訊號處理有效率地將影像中的可用特徵量化,並將實驗成果與過往研究相比較。 Online musical performances nowadays are popular as the online platforms, like Youtube or Yoku, are widely used. One specific type of performance among these is the instrument solo. Instruments solo using piano, flute, or guitar are recorded and being uploaded to the internet. Some are made as tutorial videos to teach the audiences to learn by themselves. In this paper we explore the possibility transcribing these tutorial video into readable sheet using current software technologies from signal processing and image processing, brining an improvement to the old-fashioned audio signal converting. The experiment is built upon piano solo-playing, as these kinds of performances can be found commonly on Youtube and are easier to perform the video segmentation. In the experiment the comparison between mutual information converting, the sound and video, and single information converting, which is the sound, are discussed. After the discussion, we come to a conclusion whether the adoption of correlative converting is applicable on the other instruments. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/67155 |
DOI: | 10.6342/NTU201702873 |
Fulltext Rights: | 有償授權 |
Appears in Collections: | 工程科學及海洋工程學系 |
Files in This Item:
File | Size | Format | |
---|---|---|---|
ntu-106-1.pdf Restricted Access | 2.33 MB | Adobe PDF |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.