Convolution-based classification of audio and symbolic representations of music (original) (raw)
Related papers
A Machine Learning Approach to Musically Meaningful Homogeneous Style Classification
Proceedings of AAAI, 2014
Recent literature has demonstrated the difficulty of classifying between composers who write in extremely similar styles (homogeneous style). Additionally, machine learning studies in this field have been exclusively of technical import with little musicological interpretability or significance. We present a system which addresses the difficulty of differentiating between stylistically homogeneous composers, specifically Haydn and Mozart, using best practices from supervised machine learning and musicology. Our work expands on previous style classification studies by developing more complex features as well as introducing a new class of musical features which focus on local nuances within musical scores. This system can yield interpretable musicological conclusions about Haydn’s and Mozart’s stylistic differences while distinguishing between the composers with higher accuracy than previous studies in this domain.
Composer Classification Models for Music-Theory Building
Computational Music Analysis, 2015
The task of recognizing a composer by listening to a musical piece used to be reserved for experts in music theory. The problems we address here are, first, that of constructing an automatic system that is able to distinguish between music written by different composers; and, second, identifying the musical properties that are important for this task. We take a data-driven approach by scanning a large database of existing music and develop five types of classification model that can accurately discriminate between three composers (Bach, Haydn and Beethoven). More comprehensible models, such as decision trees and rulesets, are built, as well as black-box models such as support vector machines. Models of the first type offer important insights into the differences between composer styles, while those of the second type provide a performance benchmark.
A Prototype for Classification of Classical Music using Neural Networks
As a result of recent technological innovations, there has been a tremendous growth in the Electronic Music Distribution industry. In this way, tasks such us automatic music genre classification address new and exciting research challenges. Automatic music genre recognition involves issues like feature extraction and development of classifiers using the obtained features. As for feature extraction, we use features such as the number of zero crossings, loudness, spectral centroid, bandwidth and uniformity. These are statistically manipulated, making a total of 40 features. As for the task of genre modeling, we train a feedforward neural network (FFNN). A taxonomy of subgenres of classical music is used. We consider three classification problems: in the first one, we aim at discriminating between music for flute, piano and violin; in the second problem, we distinguish choral music from opera; finally, in the third one, we aim at discriminating between all five genres. Preliminary results are presented and discussed, which show that the presented methodology may be a good starting point for addressing more challenging tasks, such as using a broader range of musical categories.
Composer Recognition Based on 2D-Filtered Piano-Rolls
2016
We propose a method for music classification based on the use of convolutional models on symbolic pitch–time representations (i.e. piano-rolls) which we apply to composer recognition. An excerpt of a piece to be classified is first sampled to a 2D pitch–time representation which is then subjected to various transformations, including convolution with predefined filters (Morlet or Gaussian) and classified by means of support vector machines. We combine classifiers based on different pitch representations (MIDI and morphetic pitch) and different filter types and configurations. The method does not require parsing of the music into separate voices, or extraction of any other predefined features prior to processing; instead it is based on the analysis of texture in a 2D pitch–time representation. We show that filtering significantly improves recognition and that the method proves robust to encoding, transposition and amount of information. On discriminating between Haydn and Mozart stri...
Performer Identification From Symbolic Representation of Music Using Statistical Models
arXiv (Cornell University), 2021
Music Performers have their own idiosyncratic way of interpreting a musical piece. A group of skilled performers playing the same piece of music would likely to inject their unique artistic styles in their performances. The variations of the tempo, timing, dynamics, articulation etc. from the actual notated music are what make the performers unique in their performances. This study presents a dataset consisting of four movements of Schubert's "Sonata in B-flat major, D.960" performed by nine virtuoso pianists individually. We proposed and extracted a set of expressive features that are able to capture the characteristics of an individual performer's style. We then present a performer identification method based on the similarity of feature distribution, given a set of piano performances. The identification is done considering each feature individually as well as a fusion of the features. Results show that the proposed method achieved a precision of 0.903 using fusion features. Moreover, the onset time deviation feature shows promising result when considered individually.
Timbre-Invariant Audio Features for Style Analysis of Classical Music
We propose a novel set of chroma-based audio features inspired by pitch class set theory and show their utility for style analysis of classical music by using them to classify recordings into historical periods. Musicologists have long studied how composers' styles develop and influence each other, but usually based on manual analyses of the score or, more recently, automatic analyses on symbolic data, both largely independent from timbre. Here, we investigate whether such musical style analyses can be realised using audio features. Based on chroma, our features describe the use of intervals and triads on multiple time scales. To test the efficacy of this approach we use a 1600 track balanced corpus that covers the Baroque, Classical, Romantic and Modern eras, and calculate features based on four different chroma extractors and several parameter configurations. Using Linear Discriminant Analysis, our features allow for a visual separation of the four eras that is invariant to timbre. Classification using Support Vector Machines shows that a high era classification accuracy can be achieved despite strong timbral variation (piano vs. orchestra) within eras. Under the optimal parameter configuration, the classifier achieves accuracies of 82.5%.
HIPI: A Hierarchical Performer Identification Model Based on Symbolic Representation of Music
2023
Automatic Performer Identification from the symbolic representation of music has been a challenging topic in Music Information Retrieval (MIR).In this study, we apply a Recurrent Neural Network (RNN) model to classify the most likely music performers from their interpretative styles. We study different expressive parameters and investigate how to quantify these parameters for the exceptionally challenging task of performer identification. We encode performerstyle information using a Hierarchical Attention Network (HAN) architecture, based on the notion that traditional western music has a hierarchical structure (note, beat, measure, phrase level etc.). In addition, we present a large-scale dataset consisting of six virtuoso pianists performing the same set of compositions. The experimental results show that our model outperforms the baseline models with an F1-score of 0.845 and demonstrates the significance of the attention mechanism for understanding different performance styles.
Classification of Recorded Classical Music Using Neural Networks
2004
As a result of recent technological innovations, there has been a tremendous growth in the Electronic Music Distribution industry. In this way, tasks such us automatic music genre classification address new and exciting research challenges. Automatic music genre recognition involves issues like feature extraction and development of classifiers using the obtained features. As for feature extraction, we use the number of zero crossings, loudness, spectral centroid, bandwidth and uniformity. These features are statistically manipulated, making a total of 40 features. Regarding the task of genre modeling, we train a feedforward neural network (FFNN) with the Levenberg-Marquardt algorithm. A taxonomy of subgenres of classical music is used. We consider three classification problems: in the first one, we aim at discriminating between music for flute, piano and violin; in the second problem, we distinguish choral music from opera; finally, in the third one, we aim at discriminating between all the abovementioned five genres together. We obtained 85% classification accuracy in the three-class problem, 90% in the two-class problem and 76% in the five-class problem. These results are encouraging and show that the presented methodology may be a good starting point for addressing more challenging tasks.
As a result of recent technological innovations, there has been a tremendous growth in the Electronic Music Distribution industry. Consequently, tasks such as automatic music genre classification address new and exciting research challenges. Automatic music genre recognition involves is sues like feature extraction and development of classifiers using the obtained features. We use the number of zero crossings, loudness, spectral centroid, bandwidth and uniformity for feature extraction. These features are statistically manipulated, making a total of 40 features. Regarding the task of genre modeling, we follow three approaches: the K-Nearest Neighbors (KNN) classifier, Gaussian Mixture Models (GMM) and feedforward neural networks (FFNN). A taxonomy of sub-genres of classical music is used. We consider three classification problems: in the first one, we aim at discriminating between music for flute, piano and violin; in the second problem, we distinguish choral music from opera; finally, in the third one, we seek to discriminate between al l five genres. The best results were obtained using FFNNs: 85% classification accuracy in the three-class problem, 90% in the two-class problem and 76% in the five-class problem. These results are encouraging and show that the presented methodology may be a good starting point for addressing more challenging tasks.
A., “Classification of Recorded Classical Music: A Methodology and a Comparative Study
As a result of recent technological innovations, there has been a tremendous growth in the Electronic Music Distribution industry. Consequently, tasks such as automatic music genre classification address new and exciting research challenges. Automatic music genre recognition involves is sues like feature extraction and development of classifiers using the obtained features. We use the number of zero crossings, loudness, spectral centroid, bandwidth and uniformity for feature extraction. These features are statistically manipulated, making a total of 40 features. Regarding the task of genre modeling, we follow three approaches: the K-Nearest Neighbors (KNN) classifier, Gaussian Mixture Models (GMM) and feedforward neural networks (FFNN). A taxonomy of sub-genres of classical music is used. We consider three classification problems: in the first one, we aim at discriminating between music for flute, piano and violin; in the second problem, we distinguish choral music from opera; finally, in the third one, we seek to discriminate between al l five genres. The best results were obtained using FFNNs: 85% classification accuracy in the three-class problem, 90% in the two-class problem and 76% in the five-class problem. These results are encouraging and show that the presented methodology may be a good starting point for addressing more challenging tasks.