Institutional Repository of Vilnius University
Balso ir muzikinių instrumentų atskyrimas dirbtinių neuroninių tinklų pagalba /
Abstract
dc:descriptionThe dissertation presents the problem of the separation of music sources and the ways to solve it. It discusses sound representation techniques and how their choice affects their separation. Evaluation metrics and datasets are presented. Mathematical and deep learning approaches to the problem are discussed. Two neural network architectures, a convolutional and a transformer neural network, are developed and trained on the data presented. The results of these models are compared with the state of the art models in terms of source-to-distortion ratio (SDR).
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2024
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Trinkūnas, Kristijonas,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD210648121&prefLang=en_US