Pasar al contenido principal

PRESERVACION DEL PATRIMONIO DE MUSICA ESPAÑOLA MANUSCRITA MEDIANTE TRANSCRIPCION AUTOMATICA

TIN2017-86576-R

•
Nombre agencia financiadora Agencia Estatal de Investigación
Acrónimo agencia financiadora AEI
Programa Programa Estatal de I+D+i Orientado a los Retos de la Sociedad
Convocatoria Proyectos I+D+i Retos de Investigación
Año convocatoria 2017
Plan Estatal 2017-2020 Plan Estatal de Investigación Científica y Técnica y de Innovación
Centro beneficiario UNIVERSIDAD DE ALICANTE

Resultados relacionados

Resultados totales (Incluyendo duplicados): 23
Encontrada(s) 1 página(s)

RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

End-to-End Neural Optical Music Recognition of Monophonic Scores

  • Calvo-Zaragoza, Jorge
  • Rizo, D.
[EN] Optical Music Recognition is a field of research that investigates how to computationally decode music notation from images. Despite the efforts made so far, there are hardly any complete solutions to the problem. In this work, we study the use of neural networks that work in an end-to-end manner. This is achieved by using a neural model that combines the capabilities of convolutional neural networks, which work on the input image, and recurrent neural networks, which deal with the sequential nature of the problem. Thanks to the use of the the so-called Connectionist Temporal Classification loss function, these models can be directly trained from input images accompanied by their corresponding transcripts into music symbol sequences. We also present the Printed Images of Music Staves (PrIMuS) dataset, containing more than 80,000 monodic single-staff real scores in common western notation, that is used to train and evaluate the neural approach. In our experiments, it is demonstrated that this formulation can be carried out successfully. Additionally, we study several considerations about the codification of the output musical sequences, the convergence and scalability of the neural models, as well as the ability of this approach to locate symbols in the input score., This work was supported by the Social Sciences and Humanities Research Council of Canada, and the Spanish Ministerio de Economia y Competitividad through Project HISPAMUS Ref. No. TIN2017-86576-R (supported by UE FEDER funds).
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación Artículo científico (JournalArticle)

Handwritten Music Recognition for Mensural notation with convolutional recurrent neural networks

  • Calvo-Zaragoza, Jorge
  • Toselli, Alejandro Héctor|||0000-0001-6955-9249
  • Vidal, Enrique|||0000-0003-4579-5196
[EN] Optical Music Recognition is the technology that allows computers to read music notation, which is also referred to as Handwritten Music Recognition when it is applied over handwritten notation. This technology aims at efficiently transcribing written music into a representation that can be further processed by a computer. This is of special interest to transcribe the large amount of music written in early notations, such as the Mensural notation, since they represent largely unexplored heritage for the musicological community. Traditional approaches to this problem are based on complex strategies with many explicit rules that only work for one particular type of manuscript. Machine learning approaches offer the promise of generalizable solutions, based on learning from just labelled examples. However, previous research has not achieved sufficiently acceptable results for handwritten Mensural notation. In this work we propose the use of deep neural networks, namely convolutional recurrent neural networks, which have proved effective in other similar domains such as handwritten text recognition. Our experimental results achieve, for the first time, recognition results that can be considered effective for transcribing handwritten Mensural notation, decreasing the symbol-level error rate of previous approaches from 25.7% to 7.0%. (C) 2019 Elsevier B.V. All rights reserved., First author thanks the support from the Spanish Ministry "HISPAMUS" project (TIN2017-86576-R), partially funded by the EU. The other authors were supported by the European Union's H2020 grant "Recognition and Enrichment of Archival Documents" (Ref. 674943), by the BBVA Foundacion through the 2017-2018 and 2018-2019 Digital Humanities research grants "Carabela" and "HistWeather - Dos Siglos de Datos Cilmaticos", and by EU JPICH project "HOME - History Of Medieval Europe"(Spanish PEICTI Ref. PCI2018-093122).
RUA. Repositorio Institucional de la Universidad de Alicante
Publicación Otros (Other)

Retrieving Music Semantics from Optical Music Recognition by Machine Translation

  • Thomae, Martha E.
  • Ríos-Vila, Antonio
  • Calvo-Zaragoza, Jorge
  • Rizo, David
  • Iñesta, José M.
  • Calvo-Zaragoza, Jorge|||/items/a4f3686d-f0fe-4317-a73c-cd711ceb1dde
  • Iñesta, José M.|||/items/4c7c78fb-09ed-43c3-a8ea-0c16445669d0
  • Ríos-Vila, Antonio|||/items/f47307e4-fa30-4f8d-b90d-b6d71e26d89b
  • Rizo, David|||/items/447626c0-6412-451c-8f26-b4a244418da2
In this paper, we apply machine translation techniques to solve one of the central problems in the field of optical music recognition: extracting the semantics of a sequence of music characters. So far, this problem has been approached through heuristics and grammars, which are not generalizable solutions. We borrowed the seq2seq model and the attention mechanism from machine translation to address this issue. Given its example-based learning, the model proposed is meant to apply to different notations provided there is enough training data. The model was tested on the PrIMuS dataset of common Western music notation incipits. Its performance was satisfactory for the vast majority of examples, flawlessly extracting the musical meaning of 85% of the incipits in the test set—mapping correctly series of accidentals into key signatures, pairs of digits into time signatures, combinations of digits and rests into multi-measure rests, detecting implicit accidentals, etc.
RUA. Repositorio Institucional de la Universidad de Alicante
Publicación Otros (Other)

Applying Automatic Translation for Optical Music Recognition’s Encoding Step

  • Ríos-Vila, Antonio
  • Esplà-Gomis, Miquel
  • Rizo, David
  • Ponce de León Amador, Pedro José
  • Iñesta, José M.
  • Esplà-Gomis, Miquel|||/items/6614b776-cf96-43c9-b300-de4c8cf17c43
  • Iñesta, José M.|||/items/4c7c78fb-09ed-43c3-a8ea-0c16445669d0
  • Ponce de León Amador, Pedro José|||/items/9301f92f-611c-4636-a5c2-89fcfbe35694
  • Ríos-Vila, Antonio|||/items/f47307e4-fa30-4f8d-b90d-b6d71e26d89b
  • Rizo, David|||/items/447626c0-6412-451c-8f26-b4a244418da2
Optical music recognition is a research field whose efforts have been mainly focused, due to the difficulties involved in its processes, on document and image recognition. However, there is a final step after the recognition phase that has not been properly addressed or discussed, and which is relevant to obtaining a standard digital score from the recognition process: the step of encoding data into a standard file format. In this paper, we address this task by proposing and evaluating the feasibility of using machine translation techniques, using statistical approaches and neural systems, to automatically convert the results of graphical encoding recognition into a standard semantic format, which can be exported as a digital score. We also discuss the implications, challenges and details to be taken into account when applying machine translation techniques to music languages, which are very different from natural human languages. This needs to be addressed prior to performing experiments and has not been reported in previous works. We also describe and detail experimental results, and conclude that applying machine translation techniques is a suitable solution for this task, as they have proven to obtain robust results.
RUA. Repositorio Institucional de la Universidad de Alicante
Publicación Otros (Other)

An MEI-based standard encoding for hierarchical music analyses

  • Rizo, David
  • Marsden, Alan
  • Rizo, David|||/items/447626c0-6412-451c-8f26-b4a244418da2
We propose a standard representation for hierarchical musical analyses as an extension to the Music Encoding Initiative (MEI) representation for music. Analyses of music need to be represented in digital form for the same reasons as music: preservation, sharing of data, data linking, and digital processing. Systems exist for representing sequential information, but many music analyses are hierarchical, whether represented explicitly in trees or graphs or not. Features of MEI allow the representation of an analysis to be directly associated with the elements of the music analyzed. MEI’s basis in TEI (Text Encoding Initiative), allows us to design a scheme which reuses some of the elements of TEI for the representation of trees and graphs. In order to capture both the information specific to a type of music analysis and the underlying form of an analysis as a tree or graph, we propose related “semantic” encodings, which capture the detailed information, and generic “non-semantic” encodings which expose the tree or graph structure. We illustrate this with examples of representations of a range of different kinds of analysis.
RUA. Repositorio Institucional de la Universidad de Alicante
Publicación Otros (Other)

Multi-Label Logo Classification using Convolutional Neural Networks

  • Gallego, Antonio-Javier
  • Pertusa, Antonio
  • Bernabeu, Marisa
  • Gallego, Antonio-Javier|||/items/8f865afc-8f63-4e0d-9408-bebc7ba836d0
  • Pertusa, Antonio|||/items/3bf633ff-4d49-4aaa-a004-abcf6fed3540
The classification of logos is a particular case within computer vision since they have their own characteristics. Logos can contain only text, iconic images or a combination of both, and they usually include figurative symbols designed by experts that vary substantially besides they may share the same semantics. This work presents a method for multi-label classification and retrieval of logo images. For this, Convolutional Neural Networks (CNN) are trained to classify logos from the European Union TradeMark (EUTM) dataset according to their colors, shapes, sectors and figurative designs. An auto-encoder is also trained to learn representations of the input images. Once trained, the neural codes from the last convolutional layers in the CNN and the central layer of the auto-encoder can be used to perform similarity search through kNN, allowing us to obtain the most similar logos based on their color, shape, sector, figurative elements, overall features, or a weighted combination of them provided by the user. To the best of our knowledge, this is the first multi-label classification method for logos, and the only one that allows retrieving a ranking of images with these criteria provided by the user.
RUA. Repositorio Institucional de la Universidad de Alicante
Publicación Otros (Other)

Handwritten Music Recognition for Mensural notation with convolutional recurrent neural networks

  • Calvo-Zaragoza, Jorge
  • Toselli, Alejandro H.
  • Vidal, Enrique
  • Calvo-Zaragoza, Jorge|||/items/a4f3686d-f0fe-4317-a73c-cd711ceb1dde
Optical Music Recognition is the technology that allows computers to read music notation, which is also referred to as Handwritten Music Recognition when it is applied over handwritten notation. This technology aims at efficiently transcribing written music into a representation that can be further processed by a computer. This is of special interest to transcribe the large amount of music written in early notations, such as the Mensural notation, since they represent largely unexplored heritage for the musicological community. Traditional approaches to this problem are based on complex strategies with many explicit rules that only work for one particular type of manuscript. Machine learning approaches offer the promise of generalizable solutions, based on learning from just labelled examples. However, previous research has not achieved sufficiently acceptable results for handwritten Mensural notation. In this work we propose the use of deep neural networks, namely convolutional recurrent neural networks, which have proved effective in other similar domains such as handwritten text recognition. Our experimental results achieve, for the first time, recognition results that can be considered effective for transcribing handwritten Mensural notation, decreasing the symbol-level error rate of previous approaches from 25.7% to 7.0%.
RUA. Repositorio Institucional de la Universidad de Alicante
Publicación Otros (Other)

Efficient k-nearest neighbor search based on clustering and adaptive k values

  • Gallego, Antonio-Javier
  • Rico-Juan, Juan Ramón
  • Valero-Mas, Jose J.
  • Gallego, Antonio-Javier|||/items/8f865afc-8f63-4e0d-9408-bebc7ba836d0
  • Rico-Juan, Juan Ramón|||/items/ab906fc8-ea97-4172-a38d-6eb429e9dc84
  • Valero-Mas, Jose J.|||/items/8ad448df-2d0f-41e8-8403-cd85202b7263
The k-Nearest Neighbor (kNN) algorithm is widely used in the supervised learning field and, particularly, in search and classification tasks, owing to its simplicity, competitive performance, and good statistical properties. However, its inherent inefficiency prevents its use in most modern applications due to the vast amount of data that the current technological evolution generates, being thus the optimization of kNN-based search strategies of particular interest. This paper introduces the caKD+ algorithm, which tackles this limitation by combining the use of feature learning techniques, clustering methods, adaptive search parameters per cluster, and the use of pre-calculated K-Dimensional Tree structures, and results in a highly efficient search method. This proposal has been evaluated using 10 datasets and the results show that caKD+ significantly outperforms 16 state-of-the-art efficient search methods while still depicting such an accurate performance as the one by the exhaustive kNN search.
RUA. Repositorio Institucional de la Universidad de Alicante
Publicación Otros (Other)

Extensions to rank-based prototype selection in k-Nearest Neighbour classification

  • Rico-Juan, Juan Ramón
  • Valero-Mas, Jose J.
  • Calvo-Zaragoza, Jorge
  • Calvo-Zaragoza, Jorge|||/items/a4f3686d-f0fe-4317-a73c-cd711ceb1dde
  • Rico-Juan, Juan Ramón|||/items/ab906fc8-ea97-4172-a38d-6eb429e9dc84
  • Valero-Mas, Jose J.|||/items/8ad448df-2d0f-41e8-8403-cd85202b7263
The k-nearest neighbour rule is commonly considered for classification tasks given its straightforward implementation and good performance in many applications. However, its efficiency represents an obstacle in real-case scenarios because the classification requires computing a distance to every single prototype of the training set. Prototype Selection (PS) is a typical approach to alleviate this problem, which focuses on reducing the size of the training set by selecting the most interesting prototypes. In this context, rank methods have been postulated as a good solution: following some heuristics, these methods perform an ordering of the prototypes according to their relevance in the classification task, which is then used to select the most relevant ones. This work presents a significant improvement of existing rank methods by proposing two extensions: (i) a greater robustness against noise at label level by considering the parameter ‘k’ of the classification in the selection process; and (ii) a new parameter-free rule to select the prototypes once they have been ordered. The experiments performed in different scenarios and datasets demonstrate the goodness of these extensions. Also, it is empirically proved that the new full approach is competitive with respect to existing PS algorithms.
RUA. Repositorio Institucional de la Universidad de Alicante
Publicación Otros (Other)

Encoding Polyphony from Medieval Manuscripts Notated in Mensural Notation

  • Desmond, Karen
  • Pugin, Laurent
  • Regimbal, Juliette
  • Rizo, David
  • Sapp, Craig
  • Thomae, Martha E.
  • Rizo, David|||/items/447626c0-6412-451c-8f26-b4a244418da2
This panel submission for the 2021 Music Encoding Conference brings together five short papers that focus on the making of computer-readable encodings of polyphony in the notational style – mensural notation – in which it was originally copied. Mensural notation was used in the medieval West to encode polyphony from the late thirteenth to sixteenth centuries. The Measuring Polyphony (MP) Online Editor, funded by an NEH Digital Humanities Advancement Grant, is a software that enables non-technical users to make Humdrum and MEI encodings of mensural notation, and links these encodings to digital images of the manuscripts in which these compositions were first notated. Topics explored by the authors include: the processes of, and the goals informing, the linking of manuscript images to music encodings; choices and compromises made in the development process of the MP Editor in order to facilitate its rapid deployment; and the implications of capturing dual encodings – a parts-based encoding that reflects the layout of the original source, and a score-based encoding. Having two encodings of the music data is useful for a variety of activities, including performance and analysis, but also within the editorial process, and for sharing data with other applications. The authors present two case studies that document the possibilities and potential in the interchange of music data between the MP Editor and other applications, specifically, MuRET, an optical music recognition (OMR) tool, and Humdrum analysis tools.
RUA. Repositorio Institucional de la Universidad de Alicante
Publicación Otros (Other)

Improving Convolutional Neural Networks’ Accuracy in Noisy Environments Using k-Nearest Neighbors

  • Gallego, Antonio-Javier
  • Pertusa, Antonio
  • Calvo-Zaragoza, Jorge
  • Calvo-Zaragoza, Jorge|||/items/a4f3686d-f0fe-4317-a73c-cd711ceb1dde
  • Gallego, Antonio-Javier|||/items/8f865afc-8f63-4e0d-9408-bebc7ba836d0
  • Pertusa, Antonio|||/items/3bf633ff-4d49-4aaa-a004-abcf6fed3540
We present a hybrid approach to improve the accuracy of Convolutional Neural Networks (CNN) without retraining the model. The proposed architecture replaces the softmax layer by a k-Nearest Neighbor (kNN) algorithm for inference. Although this is a common technique in transfer learning, we apply it to the same domain for which the network was trained. Previous works show that neural codes (neuron activations of the last hidden layers) can benefit from the inclusion of classifiers such as support vector machines or random forests. In this work, our proposed hybrid CNN + kNN architecture is evaluated using several image datasets, network topologies and label noise levels. The results show significant accuracy improvements in the inference stage with respect to the standard CNN with noisy labels, especially with relatively large datasets such as CIFAR100. We also verify that applying the ℓ2 norm on neural codes is statistically beneficial for this approach.
RUA. Repositorio Institucional de la Universidad de Alicante
Publicación Otros (Other)

Incremental Unsupervised Domain-Adversarial Training of Neural Networks

  • Gallego, Antonio-Javier
  • Calvo-Zaragoza, Jorge
  • Fisher, Robert B.
  • Calvo-Zaragoza, Jorge|||/items/a4f3686d-f0fe-4317-a73c-cd711ceb1dde
  • Gallego, Antonio-Javier|||/items/8f865afc-8f63-4e0d-9408-bebc7ba836d0
In the context of supervised statistical learning, it is typically assumed that the training set comes from the same distribution that draws the test samples. When this is not the case, the behavior of the learned model is unpredictable and becomes dependent upon the degree of similarity between the distribution of the training set and the distribution of the test set. One of the research topics that investigates this scenario is referred to as domain adaptation (DA). Deep neural networks brought dramatic advances in pattern recognition and that is why there have been many attempts to provide good DA algorithms for these models. Herein we take a different avenue and approach the problem from an incremental point of view, where the model is adapted to the new domain iteratively. We make use of an existing unsupervised domain-adaptation algorithm to identify the target samples on which there is greater confidence about their true label. The output of the model is analyzed in different ways to determine the candidate samples. The selected samples are then added to the source training set by self-labeling, and the process is repeated until all target samples are labeled. This approach implements a form of adversarial training in which, by moving the self-labeled samples from the target to the source set, the DA algorithm is forced to look for new features after each iteration. Our results report a clear improvement with respect to the non-incremental case in several data sets, also outperforming other state-of-the-art DA algorithms.
RUA. Repositorio Institucional de la Universidad de Alicante
Publicación Otros (Other)

Prototype generation in the string space via approximate median for data reduction in nearest neighbor classification

  • Castellanos, Francisco J.
  • Valero-Mas, Jose J.
  • Calvo-Zaragoza, Jorge
  • Calvo-Zaragoza, Jorge|||/items/a4f3686d-f0fe-4317-a73c-cd711ceb1dde
  • Castellanos, Francisco J.|||/items/6b515109-60a8-4a00-a12d-0726355ae966
  • Valero-Mas, Jose J.|||/items/8ad448df-2d0f-41e8-8403-cd85202b7263
The k-nearest neighbor (kNN) rule is one of the best-known distance-based classifiers, and is usually associated with high performance and versatility as it requires only the definition of a dissimilarity measure. Nevertheless, kNN is also coupled with low-efficiency levels since, for each new query, the algorithm must carry out an exhaustive search of the training data, and this drawback is much more relevant when considering complex structural representations, such as graphs, trees or strings, owing to the cost of the dissimilarity metrics. This issue has generally been tackled through the use of data reduction (DR) techniques, which reduce the size of the reference set, but the complexity of structural data has historically limited their application in the aforementioned scenarios. A DR algorithm denominated as reduction through homogeneous clusters (RHC) has recently been adapted to string representations but as obtaining the exact median value of a set of string data is known to be computationally difficult, its authors resorted to computing the set-median value. Under the premise that a more exact median value may be beneficial in this context, we, therefore, present a new adaptation of the RHC algorithm for string data, in which an approximate median computation is carried out. The results obtained show significant improvements when compared to those of the set-median version of the algorithm, in terms of both classification performance and reduction rates.
RUA. Repositorio Institucional de la Universidad de Alicante
Publicación Otros (Other)

End-to-End Neural Optical Music Recognition of Monophonic Scores

  • Calvo-Zaragoza, Jorge
  • Rizo, David
  • Calvo-Zaragoza, Jorge|||/items/a4f3686d-f0fe-4317-a73c-cd711ceb1dde
  • Rizo, David|||/items/447626c0-6412-451c-8f26-b4a244418da2
Optical Music Recognition is a field of research that investigates how to computationally decode music notation from images. Despite the efforts made so far, there are hardly any complete solutions to the problem. In this work, we study the use of neural networks that work in an end-to-end manner. This is achieved by using a neural model that combines the capabilities of convolutional neural networks, which work on the input image, and recurrent neural networks, which deal with the sequential nature of the problem. Thanks to the use of the the so-called Connectionist Temporal Classification loss function, these models can be directly trained from input images accompanied by their corresponding transcripts into music symbol sequences. We also present the Printed Music Scores dataset, containing more than 80,000 monodic single-staff real scores in common western notation, that is used to train and evaluate the neural approach. In our experiments, it is demonstrated that this formulation can be carried out successfully. Additionally, we study several considerations about the codification of the output musical sequences, the convergence and scalability of the neural models, as well as the ability of this approach to locate symbols in the input score.Optical Music Recognition is a field of research that investigates how to computationally decode music notation from images. Despite the efforts made so far, there are hardly any complete solutions to the problem. In this work, we study the use of neural networks that work in an end-to-end manner. This is achieved by using a neural model that combines the capabilities of convolutional neural networks, which work on the input image, and recurrent neural networks, which deal with the sequential nature of the problem. Thanks to the use of the the so-called Connectionist Temporal Classification loss function, these models can be directly trained from input images accompanied by their corresponding transcripts into music symbol sequences. We also present the Printed Music Scores dataset, containing more than 80,000 monodic single-staff real scores in common western notation, that is used to train and evaluate the neural approach. In our experiments, it is demonstrated that this formulation can be carried out successfully. Additionally, we study several considerations about the codification of the output musical sequences, the convergence and scalability of the neural models, as well as the ability of this approach to locate symbols in the input score.
RUA. Repositorio Institucional de la Universidad de Alicante
Publicación Otros (Other)

Automatic scale estimation for music score images

  • Castellanos, Francisco J.
  • Gallego, Antonio-Javier
  • Calvo-Zaragoza, Jorge
  • Calvo-Zaragoza, Jorge|||/items/a4f3686d-f0fe-4317-a73c-cd711ceb1dde
  • Castellanos, Francisco J.|||/items/6b515109-60a8-4a00-a12d-0726355ae966
  • Gallego, Antonio-Javier|||/items/8f865afc-8f63-4e0d-9408-bebc7ba836d0
Optical Music Recognition (OMR) is the research field focused on the automatic reading of music from scanned images. Its main goal is to encode the content into a digital and structured format with the advantages that this entails. This discipline is traditionally aligned to a workflow whose first step is the document analysis. This step is responsible of recognizing and detecting different sources of information—e.g. music notes, staff lines and text—to extract them and then processing automatically the content in the following steps of the workflow. One of the most difficult challenges it faces is to provide a generic solution to analyze documents with diverse resolutions. The endless number of existing music sources does not meet a standard that normalizes the data collections, giving complete freedom for a wide variety of image sizes and scales, thereby making this operation unsustainable. In the literature, this question is commonly overlooked and a uniform scale is assumed. In this paper, a machine learning-based approach to estimate the scale of music documents with respect to a reference scale is presented. Our goal is to propose a robust and generalizable method to adapt the input image to the requirements of an OMR system. For this, two goal-directed case studies are included to evaluate the proposed approach over common task within the OMR workflow, comparing the behavior with other state-of-the-art methods. Results suggest that it is necessary to perform this additional step in the first stage of the workflow to correct the scale of the input images. In addition, it is empirically demonstrated that our specialized approach is more promising than image augmentation strategies for the multi-scale challenge.
RUA. Repositorio Institucional de la Universidad de Alicante
Publicación Otros (Other)

White Mensural Manual Encoding: from Humdrum to MEI, Codificación manual mensural: de Humdrum a MEI

  • Rizo, David
  • Pascual León, Nieves
  • Sapp, Craig
  • Rizo, David|||/items/447626c0-6412-451c-8f26-b4a244418da2
The recovery of musical heritage currently necessarily involves its digitalization, not only by scanning images, but also by the encoding in computer-readable formats of the musical content described in the original manuscripts. In general, this encoding can be done using automated tools based with what is named Optical Music Recognition (OMR), or manually writing directly the corresponding computer code. The OMR technology is not mature enough yet to extract the musical content of sheet music images with enough quality, and even less from handwritten sources, so in many cases it is more efficient to encode the works manually. However, being currently MEI (Music Encoding Initiative) the most appropriate format to store the encoding, it is a totally tedious code to be manually written. Therefore, we propose a new format named **mens allowing a quick manual encoding, from which both the MEI format itself and other common representations such as Lilypond or the transcription in MusicXML can be generated. By using this approach, the antiphony Salve Regina for eight-voice choir written by Jerónimo de la Torre (1607–1673) has been successfully encoded and transcribed.
RUA. Repositorio Institucional de la Universidad de Alicante
Software de investigación Software (Software)

MuRET v1.0, Aplicación online de reconocimiento óptico de música y transcripción

  • Rizo, David
  • Calvo-Zaragoza, Jorge
  • Calvo-Zaragoza, Jorge|||/items/a4f3686d-f0fe-4317-a73c-cd711ceb1dde
  • Rizo, David|||/items/447626c0-6412-451c-8f26-b4a244418da2
MuRET es el acrónimo de “Music, recognition, encoding, and transcription”. El software que registramos consiste en una aplicación web que ayuda al usuario a convertir un conjunto de imágenes que contienen notación musical en una partitura digital codificada en formato MEI (Music Encoding Initiative). No se trata sólo de un OMR (optical music recognition) sino que incluye todas las funcionalidades necesarias para obtener una transcripción completa de una obra a partir de partes separadas en ficheros de entrada.
RUA. Repositorio Institucional de la Universidad de Alicante
Publicación Otros (Other)

Data representations for audio-to-score monophonic music transcription

  • Román, Miguel A.
  • Pertusa, Antonio
  • Calvo-Zaragoza, Jorge
  • Calvo-Zaragoza, Jorge|||/items/a4f3686d-f0fe-4317-a73c-cd711ceb1dde
  • Pertusa, Antonio|||/items/3bf633ff-4d49-4aaa-a004-abcf6fed3540
This work presents an end-to-end method based on deep neural networks for audio-to-score music transcription of monophonic excerpts. Unlike existing music transcription methods, which normally perform pitch estimation, the proposed approach is formulated as an end-to-end task that outputs a notation-level music score. Using an audio file as input, modeled as a sequence of frames, a deep neural network is trained to provide a sequence of music symbols encoding a score, including key and time signatures, barlines, notes (with their pitch spelling and duration) and rests. Our framework is based on a Convolutional Recurrent Neural Network (CRNN) with Connectionist Temporal Classification (CTC) loss function trained in an end-to-end fashion, without requiring to align the input frames with the output symbols. A total of 246,870 incipits from the Répertoire International des Sources Musicales online catalog were synthesized using different timbres and tempos to build the training data. Alternative input representations (raw audio, Short-Time Fourier Transform (STFT), log-spaced STFT and Constant-Q transform) were evaluated for this task, as well as different output representations (Plaine & Easie Code, Kern, and a purpose-designed output). Results show that it is feasible to directly infer score representations from audio files and most errors come from music notation ambiguities and metering (time signatures and barlines).
RUA. Repositorio Institucional de la Universidad de Alicante
Publicación Otros (Other)

Unsupervised neural domain adaptation for document image binarization

  • Castellanos, Francisco J.
  • Gallego, Antonio-Javier
  • Calvo-Zaragoza, Jorge
  • Calvo-Zaragoza, Jorge|||/items/a4f3686d-f0fe-4317-a73c-cd711ceb1dde
  • Castellanos, Francisco J.|||/items/6b515109-60a8-4a00-a12d-0726355ae966
  • Gallego, Antonio-Javier|||/items/8f865afc-8f63-4e0d-9408-bebc7ba836d0
Binarization is a well-known image processing task, whose objective is to separate the foreground of an image from the background. One of the many tasks for which it is useful is that of preprocessing document images in order to identify relevant information, such as text or symbols. The wide variety of document types, alphabets, and formats makes binarization challenging. There are multiple proposals with which to solve this problem, from classical manually-adjusted methods, to more recent approaches based on machine learning. The latter techniques require a large amount of training data in order to obtain good results; however, labeling a portion of each existing collection of documents is not feasible in practice. This is a common problem in supervised learning, which can be addressed by using the so-called Domain Adaptation (DA) techniques. These techniques take advantage of the knowledge learned in one domain, for which labeled data are available, to apply it to other domains for which there are no labeled data. This paper proposes a method that combines neural networks and DA in order to carry out unsupervised document binarization. However, when both the source and target domains are very similar, this adaptation could be detrimental. Our methodology, therefore, first measures the similarity between domains in an innovative manner in order to determine whether or not it is appropriate to apply the adaptation process. The results reported in the experimentation, when evaluating up to 20 possible combinations among five different domains, show that our proposal successfully deals with the binarization of new document domains without the need for labeled data.
RUA. Repositorio Institucional de la Universidad de Alicante
Publicación Otros (Other)

Music Encoding Conference Proceedings 2021, 19–22 July, 2021 University of Alicante (Spain): Onsite & Online

  • Münnich, Stefan
  • Rizo, David
  • Rizo, David|||/items/447626c0-6412-451c-8f26-b4a244418da2
Este documento incluye los artículos y pósters presentados en el Music Encoding Conference 2021 realizado en Alicante entre el 19 y el 22 de julio de 2022.
RUA. Repositorio Institucional de la Universidad de Alicante
Publicación Otros (Other)

Reconocimiento holístico de partituras musicales

  • Alfaro-Contreras, María
  • Calvo-Zaragoza, Jorge
  • Iñesta, José M.
  • Calvo-Zaragoza, Jorge|||/items/a4f3686d-f0fe-4317-a73c-cd711ceb1dde
  • Iñesta, José M.|||/items/4c7c78fb-09ed-43c3-a8ea-0c16445669d0
El reconocimiento de patrones con dependencia temporal es común en áreas como el reconocimiento del habla o el procesamiento del lenguaje natural. De manera análoga, encontramos el reconocimiento de texto o video en el campo de análisis de imágenes. Recientemente, las Redes Neuronales Recurrentes (RNN, por sus siglas en inglés) han sido ampliamente utilizadas para resolver estas tareas, arrojando buenos resultados, siguiendo un planteamiento conocido como reconocimiento holístico. Sin embargo, su aplicación en el campo del Reconocimiento Óptico de Música (OMR, por sus siglas en inglés) no es tan sencillo debido a la presencia de diferentes elementos en la misma posición horizontal, interrumpiendo así el flujo lineal temporal. En este artículo se estudia la capacidad de las RNN para aprender códigos que representan esta interrupción en partituras musicales homofónicas. Los resultados obtenidos demuestran que las formas serializadas para codificar el contenido musical propuestas son apropiadas para el OMR basado en RNN y que por tanto, merecen más estudio.
RUA. Repositorio Institucional de la Universidad de Alicante
Publicación Otros (Other)

Insights Into Efficient k-Nearest Neighbor Classification With Convolutional Neural Codes

  • Gallego, Antonio-Javier
  • Calvo-Zaragoza, Jorge
  • Rico-Juan, Juan Ramón
  • Calvo-Zaragoza, Jorge|||/items/a4f3686d-f0fe-4317-a73c-cd711ceb1dde
  • Gallego, Antonio-Javier|||/items/8f865afc-8f63-4e0d-9408-bebc7ba836d0
  • Rico-Juan, Juan Ramón|||/items/ab906fc8-ea97-4172-a38d-6eb429e9dc84
The increasing consideration of Convolutional Neural Networks (CNN) has not prevented the use of the k-Nearest Neighbor (kNN) method. In fact, a hybrid CNN-kNN approach is an interesting option in which the network specializes in feature extraction through its activations (Neural Codes), while the kNN has the advantage of performing a retrieval by means of similarity. However, this hybrid approach also has the disadvantages of the kNN search, and especially its high computational cost which is, in principle, undesirable for large-scale data. In this paper, we present the first comprehensive study of efficient kNN search algorithms using this hybrid CNN-kNN approach. This has been done by considering up to 16 different algorithms, each of which is evaluated with a different parametrization, in 7 datasets of heterogeneous composition. Our results show that no single algorithm is capable of covering all aspects, but rather that each family of algorithms is better suited to specific aspects of the problem. This signifies that Fast Similarity Search algorithms maintain their performance, but do not reduce the cost as much as the Data Reduction family does. In turn, the Approximated Similarity Search family is postulated as a good option when attempting to balance accuracy and efficiency. The experiments also suggest that considering statistical transformation algorithms such as Linear Discriminant Analysis might be useful in certain cases.
RUA. Repositorio Institucional de la Universidad de Alicante
Publicación Otros (Other)

Bounding Edit Distance for similarity-based sequence classification on Structural Pattern Recognition

  • Rico-Juan, Juan Ramón
  • Valero-Mas, Jose J.
  • Iñesta, José M.
  • Iñesta, José M.|||/items/4c7c78fb-09ed-43c3-a8ea-0c16445669d0
  • Rico-Juan, Juan Ramón|||/items/ab906fc8-ea97-4172-a38d-6eb429e9dc84
  • Valero-Mas, Jose J.|||/items/8ad448df-2d0f-41e8-8403-cd85202b7263
Pattern Recognition tasks in the structural domain generally exhibit high accuracy results, but their time efficiency is quite low. Furthermore, this low performance is more pronounced when dealing with instance-based classifiers, since, for each query, the entire corpus must be evaluated to find the closest prototype. In this work we address this efficiency issue for the Nearest Neighbor classifier when data are encoded as two-dimensional code sequences, and more precisely strings and sequences of vectors. For this, a set of bounds is proposed in the distance metric that avoid the calculation of unnecessary distances. Results obtained prove the effectiveness of the proposal as it reduces the classification time in percentages between 80% and 90% for string representations and between 60% and 80% for data codified as sequences of vectors with respect to their corresponding non-optimized version of the classifier.