基於MIDI之內涵式鋼琴音樂檢索系統

由於網際網路的成熟發展，數位多媒體的需求日益劇增，透過網路下載或分享影音資料已成為日常生活的一部分。數位影音壓縮的資料庫中快速有效地檢索出所需的檔案，已成為一個相當重要的研究議題。常見的搜尋引擎大多藉由文字輸入進行檢索，但卻有標記錯誤或模糊造成檢索結果錯誤的缺點。本論文針對電子音樂格式－MIDI，利用音樂轉譯及MIDI音符矩陣來具體實現音樂檢索系統。主要是利用音樂轉譯技術擷取鋼琴音樂的旋律資訊，建構出MIDI內建之音符矩陣，再與音樂資料庫進行旋律相似度的比對，進而達成音樂檢索的目的。實驗結果顯示，本系統在MIDI音樂檢索上有良好的辨識度，且隨著輸入片段長度的增加而提升。在僅有四小節之音樂輸入情況下，平均倒數排名值為0.563；在五小節及六小節之音樂輸入情況下，平均倒數排名值則分別提升為0.6725及0.7604。

關鍵字

內涵式音樂檢索；音樂轉譯；樂器數位介面；音符矩陣

並列摘要

Due to the rapid development of the Internet, downloading and sharing multimedia data through networks has been part of the human life. Therefore, efficient information retrieval from multimedia database becomes an important research issue. A common method of search engine is through textual label which, however, might be ambiguous for information retrieval. In our proposed system, we focus on the piano music in MIDI domain and realize the retrieval system by music transcription and MIDI note matrix. We first extract the music information by music transcription, and use it to build the MIDI note matrix for further matching to find the most similar music in the database.The experimental results show that the proposed method can achieve better precision with longer input queries.

並列關鍵字

content-based music retrieval ； music transcription ； MIDI ； note matrix

參考文獻

[5] YANG, Kai; ZHOU, Xi. MIDI-LAB, a Powerful Visual Basic Program for Creating MIDI Music. International Journal of Software Engineering & Applications, 2012, 3.4.

[9] CHUAN, Xiao. Cover song identification using an enhanced chroma over a binary classifier based similarity measurement framework. In: Systems and Informatics (ICSAI), 2012 International Conference on. IEEE, 2012. p. 2170-2176.

[11] CHENG, ZongChao; LIN, ChingShun; CHEN, YiHen. Fast music information retrieval using PAT tree based dynamic time warping. In: Information, Communications and Signal Processing (ICICS) 2011 8th International Conference on. IEEE, 2011. p. 1-5.

[13] RAVELLI, Emmanuel; RICHARD, Gaël; DAUDET, Laurent. Audio signal representations for indexing in the transform domain. Audio, Speech, and Language Processing, IEEE Transactions on, 2010, 18.3: 434-446.

[1] 張錦鴻著，基礎樂理，新版，全音樂譜出版社，民國71年。