Toward intelligent music information retrieval

2018

Music is a rich harmonic audio signal with variety of forms and musical dimensions. The huge canvas of music evolved through centuries and decades involves variety of music genres and features. Different musical data representation, storage meth- ods and feature classification approaches helps to understand diversities and dimensions in musical features.This paper covers music data representation methods, musical features, feature engineering, generation, selection and learning methodologies used for musical data with example application for query by humming. The example chosen uses generic music features considering no need of familiarity with specific music genre. De- tailed discussion is provided for feature generation process with different approaches used. These feature engineering examples in music data analytics are useful in various applications of content based music information retrieval such as query by humming, similarity of music, clustering, music plagiarism etc. Enorm...

ACM SIGMultimedia Records, 2009

Music is a complex form of communication in which both artists and cultures express their ideas and identity. When we listen to music we do not simply perceive the acoustics of the sound in a temporal pattern, but also its relationship to other sounds, songs, artists, cultures and emotions. Owing to the complex, culturally-defined distribution of acoustic and temporal patterns amongst these relationships, it is unlikely that a general audio similarity metric will be suitable as a music similarity metric. Hence, we are unlikely to be able to emulate human perception of the similarity of songs without making reference to some historical or cultural context.

2017

The performance of existing search engines for retrieval of images is facing challenges resulting in inappropriate noisy data rather than accurate information searched for. The reason for this being data retrieval methodology is mostly based on information in text form input by the user. In certain areas, human computation can give better results than machines. In the proposed work, two approaches are presented. In the first approach, Unassisted and Assisted Crowd Sourcing techniques are implemented to extract attributes for the classical music, by involving users (players) in the activity. In the second approach, signal processing is used to automatically extract relevant features from classical music. Mel Frequency Cepstral Coefficient (MFCC) is used for feature learning, which generates primary level features from the music audio input. To extract high-level features related to the target class and to enhance the primary level features, feature enhancement is done. During the lea...

2003

Content-based music genre classification is a fundamental component of music information retrieval systems and has been gaining importance and enjoying a growing amount of attention with the emergence of digital music on the Internet. Currently little work has been done on automatic music genre classification, and in addition, the reported classification accuracies are relatively low. This paper proposes a new feature extraction method for music genre classification, DWCHs 1 . DWCHs capture the local and global information of music signals simultaneously by computing histograms on their Daubechies wavelet coefficients. Effectiveness of this new feature and of previously studied features are compared using various machine learning classification algorithms, including Support Vector Machines and Linear Discriminant Analysis. It is demonstrated that the use of DWCHs significantly improves the accuracy of music genre classification.

2006

Abstract Efficient and intelligent music information retrieval is a very important topic of the 21st century. With the ultimate goal of building personal music information retrieval systems, this paper studies the problem of identifying" similar" artists using both lyrics and acoustic data. In this paper, we present a clustering algorithm that integrates features from both sources to perform bimodal learning.

Journal of Intelligent Information Systems

Increasing availability of music data via Internet evokes demand for efficient search through music files. Users' interests include melody tracking, harmonic structure analysis, timbre identification, and so on. We visualize, in an illustrative example, why content based search is needed for music data and what difficulties must be overcame to build an intelligent music information retrieval system.

Archives of Acoustics, 2008

This paper presents the main issues related to music information retrieval (MIR) domain. MIR is a multi-discipline area. Within this domain, there exists a variety of approaches to musical instrument recognition, musical phrase classification, melody classification (e.g. queryby-humming systems), rhythm retrieval, high-level-based music retrieval such as looking for emotions in music or differences in expressiveness, music search based on listeners' preferences, etc. The key-issue lies, however, in the parameterization of a musical event. In this paper some aspects related to MIR are shortly reviewed in the context of possible and current applications to this domain.

2007

Intelligent music navigation is one of the important tasks in today's music applications. In this context we propose several high-level musical similarity features that can be used in automatic music navigation, classification and recommendation. The features we propose use Continuous Wavelet-like Transform as a basic time-frequency analysis of a musical signal due to its flexibility in timefrequency resolutions. A novel 2D beat histogram is presented in the paper as a rhythmic similarity feature which is free from dependency on recording condition and does not require sophisticated adaptive algorithms of threshold finding in beat detection. This paper also describes a CWT based algorithm of multiple F0 estimation (note detection) and corresponding melodic similarity features). Evaluation of the both similarity measures is done in automatic genre classification context and playlist composition.

1999

We present a system capable of performing similarity queries against a large archive of digital music. Users are able to search for songs which "sound similar" to a given query song, thereby aiding the navigation and discovery of new music in such an archive. Our technique is based on reduction of the music data to a feature space of relatively small dimensionality (1248 feature dimensions per song); this is accomplished using a set of feature extractors which derive frequency, amplitude, and tempo data from the encoded music data. Queries are then performed using a k-nearest neighbor search in the feature space. Our system allows subsets of the feature space to be selected on a per-query basis.

… , Speech and Signal …, 2006

Signals and Communication Technology, 2017

In recent years, the revenue earned through digital music stood at a billion-dollar market and the US remained the most profitable market for Digital music. Due to the digital shift, today people have access to millions of music clips from online music applications through their smart phones. In this context, there are some issues identified between the music listeners, music search engine by querying and retrieving music clips from a large collection of music data set. Classification is one of the fundamental problems in music information retrieval (MIR). Still, there are some hurdles according to their listener's preferences regarding music collections and their categorization. In this paper, different music extraction features are addressed, which can be used in various tasks related to music classification like a listener's mood, instrument recognition, artist identification, genre, query-by-humming, and music annotation. This review illustrates various features that can be used for addressing the research challenges posed by music mining.

IEEE Access

An algorithm has been developed to find the similarity between given songs. The song pattern similarity has been determined by knowing the note structures and the fundamental frequencies of each note of the two songs, under consideration. The statistical concept namely Correlation of Coefficient is used in this work. Correlation of Coefficient is determined by applying 16 Note-Measure Method. If Correlation of Coefficient is near to 1, it indicates that the patterns of the two songs under consideration are similar. Otherwise, there exists a certain percentage of similarity only. This basic principle is used in a set of Indian Classical Music (ICM) based songs. The proposed algorithm can determine the similarity between songs, so that alternative songs in place of some well-known songs can be identified, in terms of the embedded raga patterns. A digital music library has been constructed as a part of this work. The library consists of different songs, their raga name, and their corresponding healing capabilities in terms of music therapy. The proposed work may find application in the area of music therapy. Music therapy is an area of research which is explored significantly in recent time. This work can also be exploited for developing intelligent multimedia tool that is applicable in healthcare domain. A multimedia based mobile app has been developed encapsulating the above mentioned idea that can recommend alternative or similar songs to the existing ICM based songs. This mobile app based music recommendation system may be used for different purposes including entertainment and healthcare. As a result of the applications of the proposed algorithm, similar songs in terms of raga patterns can be discovered from within the pool of a set of songs. A music recommendation system built on this algorithm can retrieve an alternative song from within the pool of songs as a replacement to a wellknown song, which otherwise may be used for a particular music therapy. Results are reported and analyzed thoroughly. Future scope of the work is outlined. INDEX TERMS Fundamental frequency measure (FFM), correlation of coefficient, computational musicology, music recommendation system (MIR), music therapy, electronic healthcare.

Lecture Notes in Computer Science, 2015

Music Information Retrieval (MIR) is an interdisciplinary research area that covers automated extraction of information from audio signals, music databases and services enabling the indexed information searching. In the early stages the primary focus of MIR was on music information through Query-by-Humming (QBH) applications, i.e. on identifying a piece of music by singing (singing/whistling), while more advanced implementations supporting Queryby-Example (QBE) searching resulted in names of audio tracks, song identification, etc. Both QBH and QBE required several steps, among others an optimized signal parametrization and the soft computing approach. Nowadays, MIR is associated with research based on the content analysis that is related to the retrieval of a musical style, genre or music referring to mood or emotions. Even though, this type of music retrieval called Query-by-Category still needs feature extraction and parametrization optimizing, but in this case search of global online music systems and services applications with their millions of users is based on statistical measures. The paper presents details concerning MIR background and answers a question concerning usage of soft computing versus statistics, namely: why and when each of them should be employed.

2017

─ With the advancement of technology in today’s era, there is an utmost need for reliable music retrieval methods in order to organize and search through the large music archives that are available on the internet. Music genre classification is the most fundamental and essential component in music information retrieval (MIR) systems. An appropriate choice of music features and classifier is a crucial task for developing an accurate and efficient contentbased classification system. In this work, a comparative analysis for four different set of features, viz. dynamic, timbretexture, pitch and tonal features along with the statistical parameters is examined based on the performance of respective feature set. The performance evaluation is carried out on GTZAN musical database by using support vector machine (SVM) as a classifier. The experimental results show that out of all four set of features, better classification accuracy of 95.77% is achieved for dynamic and timbre texture features.