Shih-Chuan Chiu

Algorithmic compositions based on discovered musical patterns

Computer music composition is the dream of computer music researchers. In this paper, a top-down approach is investigated to discover the rules of musical composition from given music objects and to create a new music object of which style is similar to the given music objects based on the discovered composition rules. The proposed approach utilizes the data mining techniques in order to discover the styled rules of music composition characterized by music structures, melody styles and motifs. A new music object is generated based on the discovered rules. To measure the effectiveness of the proposed approach in computer music composition, a method similar to the Turing test was adopted to test the differences between the machine-generated and human-composed music. Experimental results show that it is hard to distinguish between them. The other experiment showed that the style of generated music is similar to that of the given music objects.

Incremental mining of closed inter-transaction itemsets over data stream sliding windows

Mining inter-transaction association rules is one of the most interesting issues in data mining research. However, in a data stream environment the previous approaches are unable to find the result of the new-incoming data and the original database without re-computing the whole database. In this paper, we propose an incremental mining algorithm, called DSM-CITI (Data Stream Mining for Closed Inter-Transaction Itemsets), for discovering the set of all frequent inter-transaction itemsets from data streams. In the framework of DSM-CITI, a new in-memory summary data structure, ITP-tree, is developed to maintain frequent inter-transaction itemsets. Moreover, algorithm DSM-CITI is able to construct ITP-tree incrementally and uses the property to avoid unnecessary updates. Experimental studies show that the proposed algorithm is efficient and scalable for mining frequent inter-transaction itemsets over stream sliding windows.

Mining polyphonic repeating patterns from music data using bit-string based approaches

Mining repeating patterns from music data is one of the most interesting issues of multimedia data mining. However, less work are proposed for mining polyphonic repeating patterns. Hence, two efficient algorithms, A-PRPD (Apriori-based Polyphonic Repeating Pattern Discovery) and T-PRPD (Tree-based Polyphonic Repeating Pattern Discovery), are proposed to discover polyphonic repeating patterns from music data. Furthermore, a bit-string method is developed for improving the efficiency of the proposed algorithms. Experimental results show that the proposed algorithms, A-PRPD and T-PRPD, are both effective and efficient methods for mining polyphonic repeating patterns from synthetic music data and real data.

Computer Music Composition Based on Discovered Music Patterns

Computer music composition has been the dream of the computer music researcher. In this paper, we investigated the approach to discover the rules of music composition from given music objects, and automatically generate a new music object style similar to the given music objects. The proposed approach utilizes the data mining techniques to discover the rules of music composition characterized by the music properties, music structure, melody style and motif. A new music object is generated based on the discovered rules. To measure the effectiveness of proposed computer music composition approach, we adopted the method similar to the Turing test to test the discrimination between machine-generated and human-composed music. Experimental results showed that it is hard to discriminate. Another experiment showed that the style of generated music is similar to the given music objects.

A Study on Difficulty Level Recognition of Piano Sheet Music

Looking for a piano sheet music with proper difficulty for a piano learner is always an important work to his/her teacher. In the paper, we study on a new and challenging issue of recognizing the difficulty level of piano sheet music. To analyze the semantic content of music, we focus on symbolic music, i.e., sheet music or score. Specifically, difficulty level recognition is formulated as a regression problem to predict the difficulty level of piano sheet music. Since the existing symbolic music features are not able to capture the characteristics of difficulty, we propose a set of new features. To improve the performance, a feature selection approach, RReliefF, is used to select relevant features. An extensive performance study is conducted over two real datasets with different characteristics to evaluate the accuracy of the regression approach for predicting difficulty level. The best performance evaluated in terms of the R2 statistics over two datasets reaches 39.9% and 38.8%, respectively.

On processing continuous frequent K - N -match queries for dynamic data over networked data sources

Similarity search is one of the critical issues in many applications. When using all attributes of objects to determine their similarity, most prior similarity search algorithms are easily influenced by a few attributes with high dissimilarity. The frequent k-n-match query is proposed to overcome the above problem. However, the prior algorithm to process frequent k-n-match queries is designed for static data, whose attributes are fixed, and is not suitable for dynamic data. Thus, we propose in this paper two schemes to process continuous frequent k-n-match queries over dynamic data. First, the concept of safe region is proposed and four formulae are devised to compute safe regions. Then, scheme CFKNMatchAD-C is developed to speed up the process of continuous frequent k-n-match queries by utilizing safe regions to avoid unnecessary query re-evaluations. To reduce the amount of data transmitted by networked data sources, scheme CFKNMatchAD-C also uses safe regions to eliminate transmissions of unnecessary data updates which will not affect the results of queries. Moreover, for large-scale environments, we further propose scheme CFKNMatchAD-D by extending scheme CFKMatchAD-C to employ multiple servers to process continuous frequent k-n-match queries. Experimental results show that scheme CFKNMatchAD-C and scheme CFKNMatchAD-D outperform the prior algorithm in terms of average response time and the amount of produced network traffic.

Towards an automatic music arrangement framework using score reduction

Score reduction is a process that arranges music for a target instrument by reducing original music. In this study we present a music arrangement framework that uses score reduction to automatically arrange music for a target instrument. The original music is first analyzed to determine the type of arrangement element of each section, then the phrases are identified and each is assigned a utility according to its type of arrangement element. For a set of utility-assigned phrases, we transform the music arrangement into an optimization problem and propose a phrase selection algorithm. The music is arranged by selecting appropriate phrases satisfying the playability constraints of a target instrument. Using the proposed framework, we implement a music arrangement system for the piano. An approach similar to Turing test is used to evaluate the quality of the music arranged by our system. The experiment results show that our system is able to create viable music for the piano.

Automatic System for the Arrangement of Piano Reductions

Piano reduction is a process that arranges music for the piano by reducing the original music into the most basic components. In this study we present an automatic arrangement system for piano reduction that arranges music algorithmically for the piano while considering various roles of the piano in music. We achieve this by first analyzing the original music in order to determine the type of arrangement element performed by an instrument. Then each phrase is identified and is associated with a weighted importance value. At last, a phrase selection algorithm is proposed to select phrases with maximum importance to arrangement under the constraint of piano playability. Our experiments demonstrate that the proposed system has the ability to create piano arrangement.

On Mining Repeating Pattern with Gap Constraint

We in this paper propose a new concept, repeating patterns with gap constraint, to make repeating patterns tolerate the delay of events. To mine repeating patterns with gap constraint, we first show the anti-monotonic property of repeating patterns with gap constraint and then propose a level-wise algorithm, named G-Apriori (standing for Gap with Apriori), based on the anti-monotonic property. Similar to other level-wise mining algorithms such as Apriori, algorithm G-Apriori will scan databases several times to count the number of occurrences of each candidate repeating pattern. Such phenomenon makes G-Apriori spend much time in disk I/O, thereby making G-Apriori not suitable for large databases. In view of this, we develop an index structure to record the positions of the occurrences of each repeating pattern, and then propose algorithm GwIApriori (standing for Gap with Index Apriori) to utilize the index structure to reduce the number of database scans when mining repeating patterns with gap constraint. The experimental results show that algorithm GwI-Apriori is more scalable than algorithm G-Apriori in terms of execution time.