查詢結果分析
來源資料
頁籤選單縮合
題 名 | 流行音樂之歌唱語言辨識研究=A Study of Sung Language Identification for Popular Music |
---|---|
作 者 | 蔡偉和; 高永安; | 書刊名 | 臺北科技大學學報 |
卷 期 | 40:2 2007.11[民96.11] |
頁 次 | 頁27-40 |
分類號 | 312.85 |
關鍵詞 | 歌唱語言辨識; 音樂資訊檢索; 伴奏歌聲; 清唱訊號; 碼書; Sung language identification; Music information retrieval; Accompanied voice; Solo voice; Codebook; Phonotactic modeling; |
語 文 | 中文(Chinese) |
中文摘要 | 自動分析流行音樂之歌聲訊號是內涵式音樂資訊檢索研究領域中不可或缺且深具挑戰的工作。在本研究中,我們探討一項重要卻仍未被深入研究的問題--「歌唱語言辨識」。其目標在於達成自動識別音樂內含歌聲之演唱語言,亦即判斷歌曲所屬語言為何,例如中文歌、英文歌、或日文歌等。由於流行歌曲大都含有背景音樂伴奏,直接應用現有基於「語音辨認」之「口說語言辨識」技術於歌唱語言辨識問題將不可行。有鑑於此,本研究發展一套抽取伴奏歌聲之基本發音單元的方法,利用各語言擁有其獨特之基本發音單元組合的特性來識別語言。為達到此一目標,我們開發了(1)歌聲/純伴奏切割技術以針對含有歌聲之音樂區段進行語言辨識、(2)清唱訊號碼書求取技術以將基本發音單元表示成不需要語音辨認即可獲得的碼字、以及(3)描述歌唱中各基本發音單元之出現機率與彼此左右相鄰關係之統計模型與辨識技術。 |
英文摘要 | Automatic analysis of singing from popular music is an important and challenging issue within the research target of content-based music information retrieval. As part of this research target, this proposal presents a first attempt to automatically identify the language sung in a music recording. As a vast majority of popular music contain background accompaniment during most or all vocal passages, it is infeasible to apply the existing methods of spoken language identification based on speech recognition to the problem of sung language identification. Recognizing this, we propose to develop techniques for extracting the basic phonological units from accompanied singing voices, thereby exploiting the phonotactic characteristics of each language embedding in the combinations of the basic phonological units to distinguish one language from another. Specific work includes (1) vocal/non-vocal segmentation for locating the segments containing voices; (2) generation of solo voice codebook for deriving the basic phonological units in an unsupervised manner; (3) phonotactic modeling for characterizing both the static and dynamic information of the basic phonological units within a music recording, thereby identifying the language to which the recording belong; and (4) music database collection, annotation, and analysis. |
本系統中英文摘要資訊取自各篇刊載內容。