Similarity Estimation for Classical Indian Music

In this paper I argue that music recommendation algorithms are a complex element of contemporary digital culture. We trust music streaming and recommender systems like Spotify to ‘set the mood’ for us, to soundtrack our private lives and activities, to recommend & discover for us. These systems purport to ‘know’ us (alongside the millions of other users), and as such we let them into our most intimate listening spaces and moments. We fetishise and share the datafication of our listening habits, reflected to us annually in Spotify’s “Your 2018 Wrapped” and every Monday in ‘Discover Weekly’, even daily in the “playlists made for you”. As the accuracy of these recommendations increases, so too does our trust in these systems. ‘Bad’ or inaccurate recommendations feel like a betrayal, giving us the sense that the algorithms don’t really know us at all. Users speak of ‘their’ algorithm, as if it belonged to them and not a part of a complex machine learning recommendation system. This paper builds on research which critically examined the music recommendation system that powers Spotify and its many discovery features. The research explored the process through which Spotify automates discovery by incorporating established methods of music consumption, and demonstrated that music recommendation systems such as Spotify are emblematic of the politics of algorithmic culture.

Music Genre Classification (MGC) automatically categorizes music into different genres based on various musical attributes and features in a small number of music files. This is a crucial problem in the field of music information retrieval as it provides a way to organize and analyse large amounts of music files. MGC can be performed using conventional machine learning algorithms such as SVM, k-nearest neighbours, Decision trees, and neural networks. These algorithms learn to recognize different musical features and attributes to categorize the music files into different genres. The literature shows that the performance of conventional machine learning algorithms is inferior to deep learning algorithms such as CNN, RNN, etc., in various applications. Hence, the CNN algorithm is adapted to implement the classification of music files. This aims to classify music genres using CNN deep learning techniques. The performance of the algorithms for MGC can be evaluated using metrics such as accuracy, precision, recall, and F1-score. Additionally, the impact of different features and algorithms on the performance of MGC can be studied and compared. It has applications in areas such as automated music recommendation systems, music education, and music production. An accuracy of 83% is achieved by using CNN to accomplish the task of MGC.

The field of Music Information Retrieval is still very immature although it has been addressed by a substantial number of research projects over past three decades. Many problems are faced in music information retrieval due to the nature of music itself. In this paper, the basic representation of audio and video are discussed along with the challenges of music retrieval. The need for music information is discussed using the statistical survey. The paper challenges the common assumption that searching on the pitch alone is satisfactory and discusses the other features which can aid in searching. This assumption may indeed be right for monophonic music but it is insufficient for polyphonic music. Even in monophonic music the possibility of faulty results can’t be ignored.

Over the years, Music Information Retrieval (MIR) has proposed various models pretrained on large amounts of music data. Transfer learning showcases the proven effectiveness of pretrained backend models with a broad spectrum of downstream tasks, including auto-tagging and genre classification. However, MIR papers generally do not explore the efficiency of pretrained models for Music Recommender Systems (MRS). In addition, the Recommender Systems community tends to favour traditional end-to-end neural network learning over these models. Our research addresses this gap and evaluates the applicability of six pretrained backend models (MusicFM, Music2Vec, MERT, EncodecMAE, Jukebox, and MusiCNN) in the context of MRS. We assess their performance using three recommendation models: K-nearest neighbours (KNN), shallow neural network, and BERT4Rec. Our findings suggest that pretrained audio representations exhibit significant performance variability between traditional MIR tasks and MRS, indicating that valuable aspects of musical information captured by backend models may differ depending on the task. This study establishes a foundation for further exploration of pretrained audio representations to enhance music recommendation systems.

Melody extraction from polyphonic audio, which consists of number of different instruments, is one of the most challenging tasks in the field of Music Information Retrieval (MIR). This paper aims at implementation of extracting melody using two-way mismatch dual pitch tracking (TWMDPT) and harmonic cluster tracking (HCT) algorithms. In this work, melody of singing voice is estimated in presence of multi-accompaniment. In the first method temporal features of voice harmonics are used to find the voice pitch and it is based on salience function. The second method depends upon strong higher harmonics for robustness against distortion by the first harmonic due to low frequency accompaniments. Performance of these algorithms is evaluated on MIR-1k database and accuracy is estimated using Raw Pitch Accuracy (RPA), Raw Chroma Accuracy (RCA) and timbre feature.

A frequent problem when dealing with audio classification tasks is the scarcity of suitable training data. This work investigates ways of mitigating this problem by applying transfer learning techniques to neural network architectures for several classification tasks from the field of Music Information Retrieval (MIR). First, three state-of-the-art architectures are trained and evaluated with several datasets for the task of speech/music classification. Second, feature representations or embeddings are extracted from the trained networks to classify new tasks with unseen data. The effect of pre-training with respect to the similarity of the source and target tasks are investigated in the context of transfer learning, as well as different fine-tuning strategies.

There are two important research topics in the field of Music Information Retrieval (MIR). One is how to improve the robustness of features and the other is how to speed up the retrieval process. This paper improved the algorithms which proposed by Shazam company in these two aspects. We improve the robustness of the system by a new audio finger-printing extraction using computer graphics, and the system can recognize the recordings which get in complex environment accurately. On the other hand, we propose a recursive search algorithm based on the confidence measure to improve the retrieval speed. Quantitative analysis of the opposite experiment verifies the improvement in the retrieval speed and accuracy.

Technology has developed a lot over the last decades and has made a profound impact in almost every field. The field of Music Information Retrieval (MIR) has not been an exception to this as well, one of its most promising applications being Automatic Music Transcription (AMT). It is important to identify the active regions of various Instruments in a piece before transcription and the challenge elevates even more when the audio clips are contaminated with noise. MISNA (Musical Instrument Segregation from Noisy Clips) is a system proposed towards the identification of isolated Instruments from noisy clips which can aid towards AMT in noisy environments. The system works using statistical features (LPCC-S) derived from raw Linear Predictive Cepstral Coefficient values on very short clips of lengths 1 and 2 seconds. The system has been tested for various SNR scenarios and highest accuracies of 98.63% and 97.42% for Individual Instruments and Instrument Family identification has been obtained with the aid of Extreme Learning based classifier for a highest of 2626 clips.

Microblogs are rich sources of information because they provide platforms for users to share their thoughts, news, information, activities, and so on. Twitter is one of the most popular microblogs. Twitter users often use hashtags to mark specific topics and to link them with related tweets. In this study, we investigate the relationship between the music listening behaviors of Twitter users and a popular music ranking service by comparing information extracted from tweets with music-related hashtags and the Billboard chart. We collect users' music listening behavior from Twitter using music-related hashtags (e.g., #nowplaying). We then build a predictive model to forecast the Billboard rankings and hit music. The results show that the numbers of daily tweets about a specific song and artist can be effectively used to predict Billboard rankings and hits. This research suggests that users' music listening behavior on Twitter is highly correlated with general music trends and could play an important role in understanding consumers' music consumption patterns. In addition, we believe that Twitter users' music listening behavior can be applied in the field of Music Information Retrieval (MIR).

In this paper, we propose a unified approach to fast index-based music recognition. As an important area within the field of music information retrieval (MIR), the goal of music recognition is, given a database of musical pieces and a query document, to locate all occurrences of that document within the database, up to certain possible errors. In particular, the identification of the query with regard to the database becomes possible. The approach presented in this paper is based on a general algorithmic framework for searching complex patterns of objects in large databases. We describe how this approach may be applied to two important music recognition tasks: The polyphonic (musical score-based) search in polyphonic score data and the identification of pulse-code modulation audio material from a given acoustic waveform. We give an overview on the various aspects of our technology including fault-tolerant search methods. Several areas of application are suggested. We describe several prototypic systems we have developed for those applications including the notify! and the audentify! systems for score- and waveform-based music recognition, respectively.

Classification of musical audio signals according to expressed mood or emotion has evident applications to content-based music retrieval in large databases. Wrapper selection is a dimension reduction method that has been proposed for improving classification performance. However, the technique is prone to lead to overfitting of the training data, which decreases the generalizability of the obtained results. We claim that previous attempts to apply wrapper selection in the field of music information retrieval (MIR) have led to disputable conclusions about the used methods due to inadequate analysis frameworks, indicative of overfitting, and biased results. This paper presents a framework based on cross-indexing for obtaining realistic performance estimate of wrapper selection by taking into account the simplicity and generalizability of the classification models. The framework is applied on sets of film soundtrack excerpts that are consensually associated with particular basic emotions, comparing Naive Bayes, k-NN, and SVM classifiers using both forward selection (FS) and backward elimination (BE). K-NN with BE yields the most promising results—56.5% accuracy with only four features. The most useful feature subset for k-NN contains mode majorness and key clarity, combined with dynamical, rhythmical, and structural features.

The most common way to listen to recorded music nowadays is via streaming platforms, which provide access to tens of millions of tracks. To assist users in effectively browsing these large catalogs, the integration of music recommender systems (MRSs) has become essential. Current real‐world MRSs are often quite complex and optimized for recommendation accuracy. They combine several building blocks based on collaborative filtering and content‐based recommendation. This complexity can hinder the ability to explain recommendations to end users, which is particularly important for recommendations perceived as unexpected or inappropriate. While pure recommendation performance often correlates with user satisfaction, explainability has a positive impact on other factors such as trust and forgiveness, which are ultimately essential to maintain user loyalty.In this article, we discuss how explainability can be addressed in the context of MRSs. We provide perspectives on how explainability could improve music recommendation algorithms and enhance user experience. First, we review common dimensions and goals of recommenders explainability and in general of eXplainable Artificial Intelligence (XAI), and elaborate on the extent to which these apply—or need to be adapted—to the specific characteristics of music consumption and recommendation. Then, we show how explainability components can be integrated within a MRS and in what form explanations can be provided. Since the evaluation of explanation quality is decoupled from pure accuracy‐based evaluation criteria, we also discuss requirements and strategies for evaluating explanations of music recommendations. Finally, we describe the current challenges for introducing explainability within a large‐scale industrial MRS and provide research perspectives.

The main objective of the project is to build a system that allows users to predict genre as well as language out of some music . The Automated determination of musical genre and linguistic categorization is a machine learning based prediction model which is used to predict the genres and languages of an audio . Musical genres are categorical labels created by humans to characterize piece of music. A musical genre is characterized by common characteristics typically related to the instrumentation, rhythmic structure and harmonic content of the music. The performance and relative importance of the proposed features are investigated by training the statistical pattern recognition classifiers using real world audio collections and using machine learning algorithm.In this project, we built systems called music genre and language classification and the effects of different feature extraction methods and their various combinations were also investigated. This task is of great interest in music information retrieval and recommendation systems. In recent years, several machine learning algorithms have been developed to tackle this task, using features such spectral, temporal, and tonal characteristics of audio signals. Music genre prediction and language classification is a task that involves automatically predicting a piece of music into a specific genre category as well as language based on its audio features. Keywords-machine-learning,music,music genre and language classification

Quantification of stylistic differences between musical artists is of academic interest to the music community, and is also useful for other applications such as music information retrieval and recommendation systems. Information about stylistic differences can be obtained by comparing the performances of different artists across common musical pieces. In this article, we develop a statistical methodology for identifying and quantifying systematic stylistic differences among artists that are consistent across audio recordings of a common set of pieces, in terms of several musical features. Our focus is on a comparison of ten different orchestras, based on data from audio recordings of the nine Beethoven symphonies. As generative or fully parametric models of raw audio data can be highly complex, and more complex than necessary for our goal of identifying differences between orchestras, we propose to reduce the data from a set of audio recordings down to pairwise distances between orchestras, based on different musical characteristics of the recordings, such as tempo, dynamics, and timbre. For each of these characteristics, we obtain multiple pairwise distance matrices, one for each movement of each symphony. We develop a hierarchical multidimensional scaling (HMDS) model to identify and quantify systematic differences between orchestras in terms of these three musical characteristics, and interpret the results in the context of known qualitative information about the orchestras. This methodology is able to recover several expected systematic similarities between orchestras, as well as to identify some more novel results. For example, we find that modern recordings exhibit a high degree of similarity to each other, as compared to older recordings.

Recently, the turn volume of music data on the Internet has increased rapidly. This has increased the user's cost to find music data suiting their preference from such a large data set. We propose a content-based music search and recommendation system. This system has an interface for searching and finding music data and an interface for editing a user profile which is necessary for music recommendation. By exploiting the visualization of the feature space of music and the visualization of the user profile, the user can search music data and edit the user profile. Furthermore, by exploiting the infomation which can be acquired from each visualized object in a mutually complementary manner, we make it easier for the user to search music data and edit the user profile. Concretely, the system gives to the user an information obtained from the user profile when searching music data and an information obtained from the feature space of music when editing the user profile.