• JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
 
  Bookmark and Share
 
 
Master's Dissertation
DOI
https://doi.org/10.11606/D.45.2019.tde-23122020-172756
Document
Author
Full name
Fábio Gorodscy
E-mail
Institute/School/College
Knowledge Area
Date of Defense
Published
São Paulo, 2019
Supervisor
Committee
Queiroz, Marcelo Gomes de (President)
Faria, Regis Rossi Alves
Manzolli, Jonatas
Title in Portuguese
Estratégias de alinhamento para consulta cantarolada
Keywords in Portuguese
Consulta cantarolada
Processamento de sinais
Recuperação de informação musical
Abstract in Portuguese
Consulta cantarolada é um tópico comum em recuperação de informação musical. Na tarefa deconsulta cantarolada, uma gravação cantarolada, representando imprecisamente uma melodia dereferência, é enviada para uma aplicação, que tem como objetivo recuperar a melodia de referência.Para realizar tal tarefa, um algoritmo precisa lidar com as diferenças entre a consulta e a melodiade referência, tanto no domínio do tempo quanto no domínio da frequência.Neste trabalho revemos técnicas aplicadas em trabalhos comerciais e acadêmicos, em parti-cular alguns algoritmos apresentados na conferência internacional de recuperação de informaçãomusical (ISMIR), bem como a aplicação comercial Soundhound. Este trabalho compara resultadosdo desempenho das técnicas estudadas utilizando um amplo banco de dados aberto de consultascantaroladas.O objetivo principal deste trabalho é comparar estratégias de alinhamento, extração de caracte-rísticas, rankeamento e processamentos para a tarefa de consulta cantarolada, destacando desafios,vantagens e qualidades de cada estratégia. Além disso, analisamos a confiabilidade de uma ferra-menta comercial e comparamos seus resultados com os de outras abordagens.
Title in English
Strategies for matching in query-by-humming
Keywords in English
Music information retrieval
Query-by-humming
Signal processing
Abstract in English
Query-by-humming is a common topic in music information retrieval. In the query-by-hummingtask a hummed record, representing imprecisely a target melody, is given to an application whichis supposed to retrieve the target melody from a dataset. Any algorithm addressing this task hasto handle the deviations between query and target melodies in both time and frequency domains.This work reviews standard techniques presented in the academic literature and in commercialapplications. Algorithms presented in the conference of the international society for music infor-mation retrieval (ISMIR) are reviewed, as well as the commercial application Soundhound, whichare explored and tested. This work compares the performance of several strategies for query-by-humming within a unified query dataset.The main goal of this work is to compare strategies of matching, feature exctration, ranking andprocessing for query-by-humming, highlighting qualities of each strategy. Also, the performance ofa commercial tool is evaluated and compared to other approaches.
 
WARNING - Viewing this document is conditioned on your acceptance of the following terms of use:
This document is only for private use for research and teaching activities. Reproduction for commercial use is forbidden. This rights cover the whole data about this document as well as its contents. Any uses or copies of this document in whole or in part must include the author's name.
Publishing Date
2021-01-20
 
WARNING: Learn what derived works are clicking here.
All rights of the thesis/dissertation are from the authors
CeTI-SC/STI
Digital Library of Theses and Dissertations of USP. Copyright © 2001-2021. All rights reserved.