歌曲中人聲與無人聲片段識別之研究

Recognition of Vocal and Non-Vocal Segments in Musical Sound Tracks

吳逸中(Yi-Chung Wu)

指導教授 ： 尤信程

國立臺北科技大學/電資學院/資訊工程系研究所/碩士(2012年)

摘要

本論文的主題是利用MFCC、LPC、和LPCC不同的特徵值擷取方法，使用到HMM(隱藏馬可夫模型)歌曲經過訓練之後，再利用訓練完的模型(Model)進行辨識。歌曲資料庫會分成二個部分，一部分歌曲用來訓練，剩下歌曲做為測試。我們比較了MFCC和LPCC計算出來的概似差值(Likelihood difference)來增加辨識率；此外，我們也計算雙聲道中的左右聲道相關性的運算看是否能辨識兩聲道間的不同來辨識人聲(Vocal)和非人聲(non-Vocal)部分。

關鍵字

梅爾倒頻譜係數 ； 線性預估係數 ； 線性預估倒頻譜係數 ； 隱藏馬可夫模型

並列摘要

In this thesis, we use MFCC, LPC, LPCC feature extractions and HMM(Hidden Markov Model) tool to do training and create a model. Then use the model to recognize the testing songs. The songs in the database will be separated into two parts, training songs and testing songs. We compare MFCC and LPCC Likelihood difference to increase the recognition rate.In addition, we tried to recognize the Vocal and Non-Vocal segments by computing correlation coefficient of left channel and right channel of the stereo songs.

並列關鍵字

MFCC ； LPC ； LPCC ； HMM

參考文獻

[8] Ling Feng, Andreas Brinch Nielsen, Lars Kai Hansen,"Vocal segment classification in popular music",Technical University of Denmark Department of Informatics and Mathematical Modelling, Proc. ISMIR 2008, pp. 121-126

[4] Shankar Vembu, Stephan Baumann, "Separation of vocals from polyphonic audio recordings", Proc. ISMIR 2005. pp. 337-344

[5] George Tzanetakis,"Song-specific bootstrapping of singing voice structure", Department of Computer Science Faculty of Engineering, University of Victoria, Proc. ICME, 2004, pp. 2027-2030

[6] Rabiner, L.R, "A tutorial on hidden markov model and selected applications in speech recognition", Proceeding of the IEEE, vol. 77, no. 2, February 1989, pp. 257-286

[7] Martin F. McKinney, Jeroen Breebaart,"Features for audio and music classification", Philips Research Laboratories, The Netherlands, Proc. ISMIR 2003,