A5017327131- Music and Audio Processing
- Music Technology and Sound Studies
- Speech and Audio Processing
- Educational Technology in Learning
- Social Sciences and Policies
- Literacy and Educational Practices
- Neuroscience and Music Perception
- Speech Recognition and Synthesis
- Music History and Culture
- Advanced Text Analysis Techniques
- Data Mining Algorithms and Applications
- Semantic Web and Ontologies
- Advanced Adaptive Filtering Techniques
- Diverse Musicological Studies
- Copyright and Intellectual Property
- Computational Physics and Python Applications
- Artificial Intelligence in Law
- Human Mobility and Location-Based Analysis
- Advanced Data Compression Techniques
- Text and Document Classification Technologies
- Animal Vocal Communication and Behavior
- Emotion and Mood Recognition
- Neural Networks and Applications
- Magnetic Field Sensors Techniques
- Water Systems and Optimization
Apple (United Kingdom)
2024
University of Bristol
2010-2018
National Institute of Advanced Industrial Science and Technology
2014-2015
Baidu (China)
2015
This document describes version 0.4.0 of librosa: a Python package for audio and music signal processing. At high level, librosa provides implementations variety common functions used throughout the field information retrieval. In this document, brief overview library's functionality is provided, along with explanations design goals, software development practices, notational conventions.
We present a new system for the harmonic analysis of popular musical audio. It is focused on chord estimation, although proposed additionally estimates key sequence and bass notes. distinct from competing approaches in two main ways. First, it makes use improved chromagram representation audio that takes human perception loudness into account. Furthermore, first joint estimation chords, keys, notes fully based machine learning, requiring no expert knowledge to tune parameters. This means...
To assess the performance of an automatic chord estimation system, reference annotations are indispensable. However, owing to complexity music and sometimes ambiguous harmonic structure polyphonic music, inherently subjective, as a result any derived accuracy estimates will be subjective well. In this paper, we investigate extent confounding effect subjectivity in annotations. Our results show that is important, they affect different types systems ways. have implications for research on...
Transcribing lyrics from musical audio is a challenging research problem which has not benefited many advances made in the related field of automatic speech recognition, owing to prevalent accompaniment and differences between spoken sung voice. However, one aspect this yet be exploited by researchers that significant portions will repeated throughout song. In paper we investigate how information can leveraged form consensus transcription with improved consistency accuracy. Our results show...
We present AutoLeadGuitar, a system for automatically generating guitar solo tablatures from an input chord and key sequence. Our generates solos in distinct musical phrases, is trained using existing digital sourced the web. When AutoLeadGuitar assigns phrase boundaries, rhythms fretboard positions within probabilistic framework, guided towards tones by two user-specified parameters (chord tone preference during at end of phrases). Furthermore, guitar-specific ornaments such as hammer-ons,...
Defining and computing distances between tree structures is a classical area of study in theoretical computer science, with practical applications the areas computational biology, information retrieval, text analysis, many others. In this paper, we focus on rooted, unordered, uniquely-labelled trees such as taxonomies other hierarchies. For these, introduce intuitive concept 'local move' operation an atomic edit tree. We then SuMoTED, new distance measure trees, defined minimal number local...
Abstract Advances in chord recognition research using machine learning are hampered by two factors: the scarcity of annotated training data, and limited complexity features models used. Both problems intertwined, as with few examples, increasing model would inevitably lead to overfitting. In this paper we develop a way address first problem exploiting annotations from online databases. We show how such annotations, despite being noisy lacking exact onset times, can be put use both during...
We present AutoGuitarTab, a system for generating realistic guitar tablature given an input symbolic chord and key sequence. Our consists of two modules: AutoRhythmGuitar AutoLeadGuitar. The first these generates rhythm tablatures which outline the sequence in particular style (using Markov chains to ensure playability) performs structural analysis produce structurally consistent composition. AutoLeadGuitar lead parts distinct musical phrases, guiding pitch classes towards tones steering...
Separating the singing from a polyphonic mixed audio signal is challenging but important task, with wide range of applications across music industry and informatics research. Various methods have been devised over years, ranging Deep Learning approaches to dedicated ad hoc solutions. In this paper, we present novel machine learning method for using Conditional Random Field (CRF) approach structured output prediction. We exploit diversity previously proposed by their predictions as input...
Machine learning methods for chord recognition have improved considerably in the past few years. However, further progress seems constrained by scarcity of training data. In this paper, we show that problem can be partially solved exploiting noisy but freely and abundantly available online resources, addition to fully labeled We use these data restrict output Viterbi algorithm, resulting significant improvements over standard decoding process.
The most significant problem faced by Machine Learningbased chord recognition systems is arguably the lack of highquality training examples. In this paper, we address leveraging availability annotations from guitarist websites. We show that such can be used as partial supervision a semi-supervised method—partial since accurate timing information lacking. A particular challenge in exploitation these data their low quality, potentially even leading to performance degradation if directly....