用條件隨機域方法預測華語流行歌曲最讓人印象深刻之的歌詞

Using Conditional Random Fields Method to Predict Most Memorable Lyric in Chinese POP Music

邱昱豪(Yu-Hao Chiu)

指導教授 ： 蔡宗翰

元智大學/資訊學院/資訊工程學系/碩士(2013年)

https://doi.org/10.6838/YZU.2013.00291

摘要

本研究將探討在音樂檢索領域當中，尚未被討論的方向，一首歌當中最讓人印象深刻的句子。我們嘗試以條件隨機域（Conditional Random Fields）統計模型，建立出兩組新的特徵值，分別為內部特徵值（Internal Feature）與外部特徵值（External Feature），作為目標歌詞預測的重要判斷依據。

本研究中我們建立一個標記系統讓使用者可以對自己已聽過且熟悉的歌曲進行標記。根據這些已經標記『最印象深刻之歌詞』的歌曲，建立不同的特徵值去做比較，並且與一般人比較容易認為的『最印象深刻之歌詞』當做基礎實驗方法來做比較。

關鍵字

歌詞分析 ； 條件隨機域

並列摘要

In our research , we proposed a novel method to predict the most memorable lyric. Our research is based on Conditional Random Field (CRF) statistics method to train a predict model.

In this model, we automatically extracted two set of features, Internal features and External features ,which help us to detect the most memorable sentence.

We create a annotate system which can annotate the most memorable sentence in user’s familiar POP song . Based on these most memorable sentence in Chinese POP song, we compare different feature, different machine learning model, to find the efficient way.

並列關鍵字

Lyric Analysis ； Conditional Random Field

參考文獻

S.O. AliandZ.F.Peynircioglu.Songs and emotions: are lyrics and melodies

P. Ekman. An argument for basic emotions. Cognition and Emotion, 6(3/4):169--200, 1992.

X. Hu and J. S. Downie: “Improving mood classification in music digital libraries by combining lyrics and audio,” In Proceedings of Joint Conference on Digital Libraries, (JCDL2010)

D. Liu, L. Lu, and H. J. Zhang. Automatic mood detection from acoustic music data. In ISMIR, 2003.

J. A. Russell. Affective space is bipolar. Journal of Personality and Social Psychology, 37 (3):345--356, 1979.