基於銜接技術之音樂改作

Concatenative Audio Music Re-composition

林映孜(Yin-Tzu Lin)

指導教授 ： 吳家麟

共同指導教授 ： 張智星(Jyh-Shing Roger Jang)

國立臺灣大學/電機資訊學院/資訊網路與多媒體研究所/博士(2015年)

https://doi.org/10.6342/NTU.2015.00618

摘要

利用既有的音訊音樂相銜接而產生新的音樂，我們稱作「基於銜接技術之音樂改作(concatenative audio music re-composition)」。本論文針對此類音樂改作發展了一系列的技術。這些改作的音樂，可以應用在個人影片或是幻燈片(slideshow)，或是不間斷的舞曲集錦。基於內容分析技術，樂理，以及心理聲學理論，我們提出了多種編作與選取素材的方式。首先我們可以依照相似性，句子結尾，或是小節的資訊來決定兩段音樂的接點。接著為了使音樂的節拍能夠順暢，我們提出以心理聲學為基礎的音樂速度調整方法。而為了處理節奏跟音量相差太多的素材，我們亦相對應的提出考慮兩倍節拍的速度調整法以及音量的正規化方法。在素材的選擇方面，我們提出了兩種選擇方式。一種是直接法，先利用成對的比較去除極端的音樂素材，接著利用接點的相似度來排序。而圖形法則是先將音樂的素材都處理成為樂句，藉著巧妙的內容分析技術，我們生成了一個我們稱之為音樂骰子圖(music dice graph)的graph。利用這張圖，我們便可提供個人化的什錦歌生成服務，依照使用者指定的條件，例如結構、一定要用的音樂素材等等，產生悅耳的什錦歌。此外，我們亦開發了可供使用者選歌、設定參數、修改接點的圖形化程式介面。實驗證明了各個步驟的有效性，呈現了方法之間的比較，並可協助使用者進行適切地參數選擇。

關鍵字

音樂編輯 ； 音樂銜接技術 ； 什錦歌

並列摘要

In this dissertation, systematic techniques have been developed for helping users to make new music by concatenating existing audio materials, i.e. concatenative audio music re-composition. The re-composed music can be used as the background music for personal films and slideshows or for non-stop dance suites. Based on the content analysis techniques, music theory, and psychoacoustics, various composition and selection schemes have studied in detail. We could locate appropriate connecting positions on the basis of similarity values, phrase boundaries or bar information. Besides, psychoacoustics-based tempo adjustment methods are used to smooth the tempo of concatenated music pieces. For cases of distinct tempo or volume, effective dual tempo adjustment and volume normalization schemes have been proposed and investigated, respectively. Two different schemes are proposed for selecting materials from music collections: The straightforward scheme filtered out unfitting clips by pair wise comparison and ordered the clips by similarity values at the found connecting points. The graph-assisted scheme, first, constructed a musical dice graph from pre-processed clips based on the results of music signal analyses. Then, with the graph, we can provide personalized medley creation service, which will generate various pleasing medleys conform to the specified conditions, such as the medley structure or must-use clips. We also provide an GUI for the users to choose music clips, specify parameters and adjust concatenation boundaries. Experiment results showed the effectiveness of individual components, comparisons among methods, and provide guidelines for users to choose parameters.

並列關鍵字

music editing ； concatenating music ； musical medley

參考文獻

[26] G. Griffin, Y. Kim, and D. Turnbull, “Beat-Sync-Mash-Coder: a Web Application for Real-Time Creation of Beat-Synchronous Music Mashups,” in Proc. ICASSP, Dallas, Texas, USA, 2010, pp. 2–5. [Online]. Available: http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=5495743

[2] M. Goto, “A Chorus Section Detection Method for Musical Audio Signals and Its Application to a Music Listening Station,” IEEE Trans. ASLP, vol. 14, no. 5, pp. 1783–1794, Sep. 2006. [Online]. Available: http://ieeexplore.ieee.org/xpl/articleDetails.jsp?reload=true&arnumber=1677997

[6] Y.-T. Lin, I.-T. Liu, J.-S. R. Jang, and J.-L. Wu, “Audio Musical Dice Game: A User-preference-aware Medley Generating System,” ACM TOMM, 2014.

[7] P. Hanna, P. Ferraro, and M. Robine, “On Optimizing the Editing Algorithms for Evaluating Similarity Between Monophonic Musical Sequences,” J. New Music Res., vol. 36, no. 4, pp. 267–279, 2007. [Online]. Available: http://www.tandfonline.com/doi/abs/10.1080/09298210801927861

[8] D. Cope, Experiments in Musical Intelligence. Madison, Wisconsin, USA: A-R Editions, 1996.