吉他效果器效果辨認與延遲估計

Effect Recognition and Delay Estimation for a Guitar Effector

蕭力維(Li-Wei Hsiao)

指導教授 ： 鄭士康

國立臺灣大學/電機資訊學院/電機工程學研究所/碩士(2010年)

https://doi.org/10.6342/NTU.2010.01996

摘要

音效效果器被大量運用在電吉他的彈奏上，電吉他的樂手在彈奏他人作品時常希望能模仿原始彈奏者的音色，並且藉由模仿音色的過程學習效果器音色的調整以創造屬於自己的音色。然而每種效果器有不同的音色，且每種效果器也有多個不同的參數需要調整，因此調整音色是個繁瑣並且需要長時間學習的一件事。本論文使用支援向量機(Support Vector Machine)讓電腦在輸入純電吉他的音訊檔後能自動辨認是否加入了破音(Distortion)效果，抑或是只加入了延遲(Delay)效果或沒加入效果器的乾淨音色(Clean Tone)。並針對延遲效果分別透過音色特徵與時域特徵使用類神經網路(Neural Network)以及自相關係數(Autocorrelation)方法自動偵測延遲效果器的參數設定並討論其結果，且發現使用自相關係數方法其三個參數之平均正確率為90.53%，相較於類神經網路有非常顯著的提升，並且有判斷是否加入延遲效果的好處，其偵測率為88.89%。

關鍵字

效果器辨認 ； 延遲參數估計 ； 支援向量機 ； 類神經網路 ； 自相關係數

並列摘要

Audio effects are commonly used in playing electric guitar. When playing others’ piece, guitar players often want to imitate the original guitar tone of the song and further create their own tone to tune process. However, because there are a lot of different effects, and there are many parameters on each effect, this tuning process is very tedious and needs to take much time for learning. In this thesis we let computer automatically recognize whether the input audio files which recorded guitar only have passed distortion effect or delay effect or neither (clean tone) based on SVM (Support Vector Machine). For delay effect, timbral features and time domain features of data for neural network and autocorrelation method can be applied in delay parameters estimation. The average accuracy of three delay parameters with autocorrelation method is 90.53%, and it’s significantly improved as opposed to neural network based method. The autocorrelation method can also detect whether the input data have passed delay effect, and the hit rate is 88.89%.

並列關鍵字

effect recognition ； delay parameters estimation ； support vector machine ； neural network ； autocorrelation

參考文獻

[2] T. Li, "Musical genre classification of audio signals," IEEE Transactions on Speech and Audio Processing, vol. 10, no. 5, pp. 293-302, 2002.

[3] C. Silla Jr, et al., "A machine learning approach to automatic music genre classification," Journal of the Brazilian Computer Society, vol. 14, pp. 7-18, 2008.

[4] P. Annesi, et al., "Audio feature engineering for automatic music genre classification," RIAO, Pittsburgh, 2007.

[6] T. Li, et al., "A comparative study on content-based music genre classification," in SIGIR'03, 2003, pp. 282-289.

[9] C. Liu and C. Huang, "A singer identification technique for content-based classification of MP3 music objects," in In Proceedings of International Conference on Information and Knowledge Management, 2002, pp. 438-445.