抽取出音樂當中的近似重複樣式來做音樂檢索可以減少要比對的資料量，但是使用者若使用沒有重複的旋律來查詢便會有找不到歌曲的情況。另一方面，將音樂分段成phrase可以減少樹狀索引結構的空間，亦可減少查詢處理時間，但是使用者的查詢若是跨越phrase的，也將影響查詢結果。 在本論文中，我們比較了以近似重複樣式與phrase兩種不同的音樂特徵用來做音樂檢索的效果以及效率。根據實驗顯示，使用者的查詢是重複旋律的機會大於單一phrase，所以用近似重複樣式作為音樂查詢比對資料效果是比phrase好的。而在1-D List索引結構下，近似重複樣式的效率也優於phrase。除此之外，本論文也提出了一個新的近似重複樣式抽取方法，實驗證明我們的方法是有效的。 / Extract the approximate repeating pattern from music data will decrease the volumes of music data that need to be tested when music retrieve. If the user’s query is not a repeating melody, it can’t retrieve the music that the user wants correctly. In addition, segment the music by phrase will decrease the space that tree-like index structure need, and also decrease the retrieval processing time. If the user’s query is not a single phrase, it will influence the effectiveness of retrieval. In this thesis, we compare the effectiveness and efficiency of music retrieval methods with two different music features (approximate repeating pattern and phrase). According to experiment results, the probability that user’s query is repeating melody is more than the probability that user’s query is a single phrase. Therefore, we are of the opinion that the effectiveness that use approximate repeating pattern to process retrieval is more prominent than the effectiveness that phrase to process retrieval. Furthermore, the efficiency that use approximate repeating pattern to process retrieval is more outstanding than use phrase under 1-D List index structure. Besides, a new approximate repeating pattern extraction method is proposed. Experiment results show that our approximate repeating pattern extraction method can work correctly.

Le travail de recherche examine plusieurs problématiques relatives à la perception, la représentation et la catégorisation des stimuli musicaux durant l’écoute. Nous souhaitons enquêter ces processus cognitifs dans le cadre des différentes approches théorétiques présentes dans la littérature scientifique internationale. En particulier, la thèse s’est focalisée sur le processus de segmentation perceptif du morceau pendant l'écoute, et a analysé au moyen de deux expériences comportementales, les différents rôles des nombreuses caractéristiques structurelles et dynamiques dans le développement de la représentation de la composition musicale par Pauditeur. Ils sont aussi considérés les variables liées au musicien et à l’écouter. Les données expérimentales obtenues sont étudiées en relation avec les modèles modernes de auditory map of salience, et parallèlement, avec les modèles plus spécifiques de segmentation développés pendant ces trente dernières années dans la cadre de la psychologie cognitive de la musique. Pour les expériences on a utilisé un paradigme de segmentation musical avec deux écoutes de morceaux atonales et un ordre balancé de présentation. Les résultats expérimentaux démontrent que la carte de saillance n'est pas une trame immuable pouvant être remplie avec des combinaisons de caractéristiques du stimulus. Au contraire, elle peut être modulée par la répartition de l'attention « goal directed » il travers, par exemple, une modulation des seuils perceptifs spécifiques pour certaines caractéristiques. / This research examines several issues related to the collection, representation and the categorization of musical stimuli during the listening. We investigate these cognitive processes in the with reference to the different theoretical approaches existing in the international scientific literature. In particular, the thesis focuses on the process of perceptual segmentation of musical pieces during the listening. Two behavioral experiments allow analyzing the different roles of many structural and dynamic features in the development of the listeners’ representation of the music. Experiments take into account also the variables related to the musician and the listener. The experimental data obtained are discussed with regard to the current models of auditory map of salience, as well as with models of music segmentation models. In the paradigm of musical segmentation we used subjects have to hear and segment two versions of an atonal piece. Order of presentation is balanced across participants. The results demonstrate that the saliency map is not an immutable frame deriving only from the features of the stimuli. On the opposite, it can be modulated by goal-directed attention through, for example, modulation of specific perceptual thresholds for certain characteristics.

Esta dissertação apresenta um estudo comparativo de diferentes métodos computacionais de segmentação estrutural musical, onde o principal objetivo é delimitar fronteiras de seções musicais em um sinal de áudio, e rotulá-las, i.e. agrupar as seções encontradas que correspondem a uma mesma parte musical. São apresentadas novas propostas para segmentação estrutural nãosupervisionada, incluindo métodos para processamento em tempo real, alcançando resultados com taxas de erro inferiores a 12%. O método utilizado compreende um estudo dos descritores sonoros e meios de modelá-los temporalmente, uma exposição das técnicas computacionais de segmentação estrutural e novos métodos de avaliação dos resultados que penalizam tanto a incorreta detecção das fronteiras quanto o número incorreto de rótulos encontrados. O desempenho de cada técnica computacional é calculado utilizando diferentes conjuntos de descritores sonoros e os resultados são apresentados e analisados tanto quantitativa quanto qualitativamente. / A comparative study of different music structural segmentation methods is presented, where the goal is to delimit the borders of musical sections and label them, i.e. group the sections that correspond to the same musical part. Novel proposals for unsupervised segmentation are presented, including methods for real-time segmentation, achieving expressive results, with error ratio less then 12%. Our method consists of a study of sound descriptors, an exposition of the computational techniques for structural segmentation and the description of the evaluation methods utilized, which penalize both incorrect boundary detection and incorrect number of labels. The performance of each technique is calculated using different sound descriptor sets and the results are presented and analysed both from quantitative and qualitative points-of-view.

近幾年來，華語流行音樂的發展越來越多元，而大眾所接收到的資訊是流行音樂當中的組成元素”曲與詞”，兩者分別具有賦予人類感知的功能，使人能夠深刻體會音樂作品當中所表答的內容與意境。然而，作曲與作詞都是屬於專業的創作藝術，作詞者通常在填詞時，會先對樂曲當中的結構進行粗略的分析，找出整首曲子的曲式，而針對可以填詞的部份，再進行更細部的分析將詞填入最適當的位置。流行音樂當中，曲與詞存在著密不可分的關係，瞭解歌曲結構不僅能降低填詞的門檻，亦能夠明白曲子的骨架與脈絡;在音樂教育與音樂檢索方面亦有幫助。 本研究的目標為，使用者輸入流行音樂歌曲，系統會自動分析出曲子的『曲式結構』。方法主要分成三個部分，分別為主旋律擷取、歌句分段與音樂曲式結構擷取。首先，我們利用Support Vector Machine以學習之方式建立模型後，擷取出符號音樂中之主旋律。第二步驟我們以”歌句”為單位，對主旋律進行分段，對於分段之結果建構出Self-Similarity Matrix矩陣。最後再利用Non-Negative Matrix Factorization針對不同特徵值矩陣進行分解並建立第二層之Self-Similarity Matrix矩陣，以歧異度之方式找出曲式邊界。 我們針對分段方式對歌曲結構之影響進行分析與觀察。實驗數據顯示，事先將歌曲以歌句單位分段之效果較未分段佳，而歌句分段之評測結果F-Score為0.82;將音樂中以不同特徵值建構之自相似度矩進行Non-Negative Matrix Factorization後，另一空間中之基底特徵更能有效地分辨出不同的歌曲結構，其F-Score為0.71。 / Music structure analysis is helpful for music information retrieval, music education and alignment between lyrics and music. This thesis investigates the techniques of music structure analysis for Chinese popular music. Our work is to analyze music form automatically by three steps, main melody finding, sentence discovery, and music form discovery. First, we extract main melody based on learning from user-labeled sample using support vector machine. Then, the boundary of music sentence is detected by two-way classification using support vector machine. To discover the music form, the sentence-based Self-Similarity Matrix is constructed for each music feature. Non-negative Matrix Factorization is employed to extract the new features and to construct the second level Self-Similarity Matrix. The checkerboard kernel correlation is utilized to find music form boundaries on the second level Self-Similarity Matrix. Experiments on eighty Chinese popular music are performed for performance evaluation of the proposed approaches. For the main melody finding, our proposed learning-based approach is better than existing methods. The proposed approaches achieve 82% F-score for sentence discovery while 71% F-score for music form discovery.

Esta dissertação apresenta um estudo comparativo de diferentes métodos computacionais de segmentação estrutural musical, onde o principal objetivo é delimitar fronteiras de seções musicais em um sinal de áudio, e rotulá-las, i.e. agrupar as seções encontradas que correspondem a uma mesma parte musical. São apresentadas novas propostas para segmentação estrutural nãosupervisionada, incluindo métodos para processamento em tempo real, alcançando resultados com taxas de erro inferiores a 12%. O método utilizado compreende um estudo dos descritores sonoros e meios de modelá-los temporalmente, uma exposição das técnicas computacionais de segmentação estrutural e novos métodos de avaliação dos resultados que penalizam tanto a incorreta detecção das fronteiras quanto o número incorreto de rótulos encontrados. O desempenho de cada técnica computacional é calculado utilizando diferentes conjuntos de descritores sonoros e os resultados são apresentados e analisados tanto quantitativa quanto qualitativamente. / A comparative study of different music structural segmentation methods is presented, where the goal is to delimit the borders of musical sections and label them, i.e. group the sections that correspond to the same musical part. Novel proposals for unsupervised segmentation are presented, including methods for real-time segmentation, achieving expressive results, with error ratio less then 12%. Our method consists of a study of sound descriptors, an exposition of the computational techniques for structural segmentation and the description of the evaluation methods utilized, which penalize both incorrect boundary detection and incorrect number of labels. The performance of each technique is calculated using different sound descriptor sets and the results are presented and analysed both from quantitative and qualitative points-of-view.

在音樂中，組曲是一種特別的創作形式。組曲將多首音樂段落組合排列，並且在音樂段落之間加入間奏，形成一首音樂組曲。組曲的編曲重點在於音樂段落的編排順序及段落之間的連結。平時在宴會、舞會、餐廳、賣場等場合中，往往都會連續播放多首流行音樂。利用電腦編曲自動產生流行音樂組曲，將可提升播放音樂的銜接與流暢感。 因此，本研究利用資料探勘技術及音樂編曲理論，將多首音樂重新改編成一首組曲。系統首先將每首音樂分段並找出每首音樂的代表段落。接著，系統根據代表段落間的相似度編排順序。最後，為了達到組曲中音樂段落連接的流暢性，我們以模型訓練的方式在段落連結間加入間奏。系統從訓練資料學習產生旋律發展、和弦進程與節奏的模型，接著分析代表段落的動機、旋律、和弦及節奏，使得組曲編曲後的段落連結更為流暢且完整。本研究以流行音樂鋼琴伴奏曲為測試資料，我們分別邀請三十四位受過音樂訓練與未受音樂訓練的測試者，針對本研究所提出的鋼琴伴奏節奏辨識、代表段落萃取、段落順序編排及間奏產生，評估其效果。實驗結果顯示，本研究所提出的順序編排與間奏產生技術，對於組曲的流暢感，有著相當大的幫助。 / In music, a medley is an organized piece composed from segments of existing pieces. Ordering and bridge for connection between segments are the important elements for medley arrangement. Automatic medley arrangement is helpful for playing a set of music continuously which is common in banquet, party, restaurant, shopping mall, etc.. This thesis aims to develop the automatic medley arrangement method by using data mining techniques and music arrangement theory. The proposed method first segments each music and discovers the significant segment from each music. Then, the linear arrangement based on the similarities between significant segments is generated. Finally, in order to connect segments smoothly in the medley, the bridge between two segments is generated and inserted by using model training. Three models, melody progression, chord progression and rhythm models are learned from training data. For the experiments, testing data is collected from popular piano music and thirty-four people are invited to evaluate the effectiveness of the rhythm recognition of accompaniment, the extraction of significant segment, the linear arrangement of segments, and the creation of bridge. Experimental results show that the proposed medley arrangement method performs well.