小提琴即時跟譜及自動伴奏系統

Real-time Score Following and Auto Accompaniment System for Violin

鄭家亮(Jia-Liang Cheang)

指導教授 ： 陳恆佑

共同指導教授 ： 張克寧(Keh-Ning Chang)

國立暨南國際大學/科技學院/資訊工程學系/碩士(2010年)

https://doi.org/10.6837/NCNU.2010.00220

摘要

對小提琴學習者而言，能夠盛裝站上有樂隊伴奏的舞台，奏出優美的樂章，是多麼榮耀的一件事。然而小提琴學習者獨自練習時，不僅沒有舞台，甚至沒有伴奏樂，或有伴奏樂但速度固定，練習過程並不那麼有趣。因此我們提出用電腦系統來自動伴奏，隨著小提琴練習者速度的變化提供對應的伴奏節奏，營造現場演奏的效果，讓小提琴的獨自練習更生動、更活潑。
系統首先使用音高辨識演算法 (pitch detection algorithm) 計算出小提琴演奏者彈奏的音高，再以樂譜追蹤及速度預測演算法 (score matching and tempo prediction algorithm) 即時顯示其對應到樂譜上的位置，並計算快慢比率，再用聲音展延演算法 (time scale modification algorithm) 調整伴奏樂的速度，以達到即時伴奏的效果。
最後，我們設計了一份問卷讓使用者對系統做評估。我們找來一些小提琴練習者(其中包括2位小提琴老師)對我們系統做出評估。評估統計顯示大部分使用者都對本系統有很大的興趣，並認為自動伴奏及跟譜系統對於小提琴教學上確實有幫助。

關鍵字

小提琴 ； 音高偵測 ； 節奏預測 ； 跟譜系統 ； 自動伴奏

並列摘要

This thesis aims to design a real-time score following and auto-accompaniment system for violin learners. The system plays two roles in violin practicing: “accompanist” and “score page turner.” As an accompanist, our system plays a prerecorded audio file at a variable rate that follows the tempo of the soloist’s acoustic signal. As a score page turner, our system shows the position of the soloist’s current playing note and automatically turns to the next page when it arrives at the end of the current page. Combining these two functions, our system provides a platform that makes the dullness in violin practice into an entertaining experience.
We use a pitch detection algorithm to analyze the soloist's acoustic signal and estimate the musical notes. Then a score matching and tempo prediction algorithm is applied to this signal to show the soloist's current position on the music score in real-time, and also calculate the percentage of the difference in speed. Finally, we use a time scale modification algorithm that stretches or shrinks the prerecorded accompaniment audio at a variable rate, to adapt to the soloist's performance, simulating a human accompanist.

並列關鍵字

Violin ； pitch detection ； tempo prediction ； score following system ； auto accompaniment

參考文獻

[Amado08] R.G. Amado and J.V. Filho, “Pitch detection algorithms based on zero-cross rate and autocorrelation function for musical notes,” Proceedings of the IEEE International Conference on Audio, Language and Image Processing, pp. 449-454, Jul 2008.

[Abe90] R. Abe and H. Kambayashi, “Frequency estimation based on harmonic product spectrum method,” Proceedings of the IEEE Transactions on Acoustics, Speech and Signal Process, vol. 38, no. 2, pp. 356-360, Feb 1990.

[Black00] T.R. Black and K.D. Donohue, “Pitch determination of music signals using the generalized spectrum,” Proceedings of the IEEE, Southeastcon, pp. 104-109, Apr 2000.

[Botros99] N.M. Botros and R.S. Adamjee, “Speech-pitch detection using maximum likelihood algorithm,” Proceedings of the First Joint BMES/EMBS Conference, vol. 2, pp. 882, Oct 1999.

[Cook95] N. Cook, “Music Minus One: Rock, Theory, and Performance,” New Formations 27, pp. 23-41, 1996.