A music retrieval system that accepts hummed tunes as queries is described in this paper. This system uses similaxity retrieval because a hummed tune may contain errors. The retrieval result is a list of song names ranked according to the closeness of the match. Our ultimate goal is that the correct song should be first on the list. This means that eventually our system's similarity retrieval should allow for only one correct answer.The most significant improvement our system has over general query-by-humming systems is that all processing of musical information is done based on beats instead of notes. This type of query processing is robust against queries generated from erroneous input. In addition, acoustic information is transcribed and converted into relative intervals and is used for making feature vectors. This increases the resolution of the retrieval system compared with other general systems, which use only pitch direction information.The database currently holds over 10,000 songs, and the retrieval time is at most one second. This level of performance is mainly achieved through the use of indices for retrieval. In this paper, we also report on the results of music analyses of the songs in the database. Based on these results, new technologies for improving retrieval accuracy, such as partial feature vectors and or'ed retrieval among multiple search keys, axe proposed. The effectiveness of these technologies is evaluated quantitatively, and it is found that the retrieval accuracy increases by more than 20% compared with the previous system [9]. Practical user interfaces for the system axe also described.
This paper describes a music retrieval system that accepts a hummed tune as a query. The system realizes fast and accurate retrieval through the use of a high-performance similarity retrieval technique for multiple high dimensional feature vectors generated from tone transition information and tone distribution information.System Advantages
IntroductionExpert systems are becoming increasingly important in industrial applications, mainly for real-time complex control systems, such as process control and network operation. In the field of real-time operation, expert systems must handle random input data such as sensor readings or alarms while processing inferences.To handle these data, it has been proposed that the inference engine and receiving process be separated. However, all input data are treated uniformly by the inference engine, so its load is still excessive [1] [2] [3][4].This paper suggests an algorithm and an inference mechanism which decrease the data processed by the inference engine in order to reduce its load.First, we clarified the scope of checked data, through data-dependency analysis. The analysis is performed by determining the sequence of inferences. Next, we describe a new inference mechanism using a multiprocess which consists of inference process (engine) and an auxiliary process (scheduler). The scheduler processes the input data, so that only events of direct relevance are passed through to the inference engine and the amount of inference processing decreases [5].In this research, we discuss inference processes for real-time control on the premise of a production system. A traditional production system consists of a knowledge base and an inference process (IE: Inference Engine) [6] [7] (Fig.l(a)). The knowledge base is a collection of rules in the form of "IF condition TIIEN action" statements. Tile inference engine consists of cycles of Pattern-Matching (MATCH), Conflict-Resolution (SELECT) and Action (ACT). The data. in local memory (WM: Working Memory) is examined by the inference engine only. For example, in diagnosing a breakdown, the cause of the breakdown is determined based on the rules and the state of the system at the point of breakdown. The Problem of Real-Time InferenceWhen we apply expert systems to the fields of realtime control, the inference engine must examine not only the data in WM but also random input data., which generally have the following characteristics [1][8]:1) Generated independently of the inference sequence 2) Values are changed frequently.If these input data are processed with the premise that changes are few, as in a traditional inference en-245
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
customersupport@researchsolutions.com
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.