Music Source Separation with Neural Networks

Student: Groshev Maksim

Faculty: HSE Tikhonov Moscow Institute of Electronics and Mathematics (MIEM HSE)

Educational Programme: Systems Analysis and Mathematical Technologies (Master)

Year of Graduation: 2024

This work is dedicated to the study of the problem of musical source separation. The task boils down to extracting individual instruments or vocal parts from mixed audio. In this work, existing solutions for this task are considered. A model was also developed that uses magnitude and phase as features, and dynamic data augmentation. Experiments were conducted to train the model. As a result, a model was obtained for separating music into 4 sources: drums, bass, other, vocals, which fully meets the technical specifications. The work contains 49 pages, 13 figures, 19 tables, 44 bibliographic references.

Student Theses at HSE must be completed in accordance with the University Rules and regulations specified by each educational programme.

Summaries of all theses must be published and made freely available on the HSE website.

The full text of a thesis can be published in open access on the HSE website only if the authoring student (copyright holder) agrees, or, if the thesis was written by a team of students, if all the co-authors (copyright holders) agree. After a thesis is published on the HSE website, it obtains the status of an online publication.

Student theses are objects of copyright and their use is subject to limitations in accordance with the Russian Federation’s law on intellectual property.

In the event that a thesis is quoted or otherwise used, reference to the author’s name and the source of quotation is required.

Search all student theses