Statistical piano reduction controlling performance difficulty

Nakamura, Eijiro; Yoshii, Kazuyoshi

doi:10.1017/atsip.2018.18

Cited by 16 publications

(9 citation statements)

References 15 publications

(45 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The present fingering model can be used as a prior model for complementing data exposed to noise and enhancing the precision of such automation techniques. The model can also be used as a prior model of piano scores for music transcription and generation tasks [24], to control the performance difficulty and induce the output scores more similar to human-composed one.…”

Section: Resultsmentioning

confidence: 99%

“…From a computational viewpoint, finding an appropriate fingering for these musical instruments often involves a complex combinatorial optimization problem. Automatic fingering estimation [1,3,8,11,13,18,21,23,28,39,41] has been a topic of music information processing with the aim of understanding the computational process of music performance and is used both in performance assistance and education systems [16,31,34,38] as well as for music arrangement [12,24]. There are also a growing number of public repositories 1 related to piano fingering generation, indicating the general interests in this topic.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Statistical learning and estimation of piano fingering

Nakamura

Saito

Yoshii

2020

Information Sciences

Self Cite

View full text Add to dashboard Cite

Automatic estimation of piano fingering is important for understanding the computational process of music performance and applicable to performance assistance and education systems. While a natural way to formulate the quality of fingerings is to construct models of the constraints/costs of performance, it is generally difficult to find appropriate parameter values for these models. Here we study an alternative data-driven approach based on statistical modeling in which the appropriateness of a given fingering is described by probabilities. Specifically, we construct two types of hidden Markov models (HMMs) and their higher-order extensions. We also study deep neural network (DNN)-based methods for comparison. Using a newly released dataset of fingering annotations, we conduct systematic evaluations of these models as well as a representative constraint-based method. We find that the methods based on high-order HMMs outperform the other methods in terms of estimation accuracies. We also quantitatively study individual difference of fingering and propose evaluation measures that can be used with multiple ground truth data. We conclude that the HMM-based methods are currently state of the art and generate acceptable fingerings in most parts and that they have certain limitations such as ignorance of phrase boundaries and interdependence of the two hands.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Statistical learning and estimation of piano fingering

Nakamura

Saito

Yoshii

2020

Information Sciences

Self Cite

View full text Add to dashboard Cite

show abstract

“…It would be interesting to investigate how the proposed method works on genres other than popular music. Another interesting possibility is to integrate language models [37, 38] and acoustic models [19, 39] that can deal with chords for polyphonic piano transcription. Eventually, based on the proposed framework, we aim to build a unified audio-to-score transcription system that can estimate musical scores of multiple parts of popular music.…”

Section: Discussionmentioning

confidence: 99%

Audio-to-score singing transcription based on a CRNN-HSMM hybrid model

Nishikimi

Nakamura

Goto

et al. 2021

SIP

Self Cite

View full text Add to dashboard Cite

This paper describes an automatic singing transcription (AST) method that estimates a human-readable musical score of a sung melody from an input music signal. Because of the considerable pitch and temporal variation of a singing voice, a naive cascading approach that estimates an F0 contour and quantizes it with estimated tatum times cannot avoid many pitch and rhythm errors. To solve this problem, we formulate a unified generative model of a music signal that consists of a semi-Markov language model representing the generative process of latent musical notes conditioned on musical keys and an acoustic model based on a convolutional recurrent neural network (CRNN) representing the generative process of an observed music signal from the notes. The resulting CRNN-HSMM hybrid model enables us to estimate the most-likely musical notes from a music signal with the Viterbi algorithm, while leveraging both the grammatical knowledge about musical notes and the expressive power of the CRNN. The experimental results showed that the proposed method outperformed the conventional state-of-the-art method and the integration of the musical language model with the acoustic model has a positive effect on the AST performance.

show abstract

“…Extending on the same idea, Ariga et al ( 2017 ) created a guitar solo generator that considers the fingerings as a way to measure and control the difficulty of the generated solos. Nakamura and Yoshii ( 2018 ) describes a system that creates piano reduction of ensemble scores, capable of generating reductions with different levels of difficulty based on fingering and tempo information.…”

Section: Open Challenges For Music Generation Systemsmentioning

confidence: 99%

Computational Creativity and Music Generation Systems: An Introduction to the State of the Art

Carnovalini

Rodà

2020

Front. Artif. Intell.

View full text Add to dashboard Cite

Computational Creativity is a multidisciplinary field that tries to obtain creative behaviors from computers. One of its most prolific subfields is that of Music Generation (also called Algorithmic Composition or Musical Metacreation), that uses computational means to compose music. Due to the multidisciplinary nature of this research field, it is sometimes hard to define precise goals and to keep track of what problems can be considered solved by state-of-the-art systems and what instead needs further developments. With this survey, we try to give a complete introduction to those who wish to explore Computational Creativity and Music Generation. To do so, we first give a picture of the research on the definition and the evaluation of creativity, both human and computational, needed to understand how computational means can be used to obtain creative behaviors and its importance within Artificial Intelligence studies. We then review the state of the art of Music Generation Systems, by citing examples for all the main approaches to music generation, and by listing the open challenges that were identified by previous reviews on the subject. For each of these challenges, we cite works that have proposed solutions, describing what still needs to be done and some possible directions for further research.

show abstract

Statistical piano reduction controlling performance difficulty

Cited by 16 publications

References 15 publications

Statistical learning and estimation of piano fingering

Statistical learning and estimation of piano fingering

Audio-to-score singing transcription based on a CRNN-HSMM hybrid model

Computational Creativity and Music Generation Systems: An Introduction to the State of the Art

Contact Info

Product

Resources

About