prikaz prve stranice dokumenta Razdvajanje instrumentala i glasa u audio datoteci
  Download
PDF 1.06 MB
undergraduate thesis
Razdvajanje instrumentala i glasa u audio datoteci
Osijek: Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, 2023. urn:nbn:hr:200:309870
Vodička, David
Josip Juraj Strossmayer University of Osijek
Faculty of Electrical Engineering, Computer Science and Information Technology Osijek
Department of Software Engineering
Chair of Visual Computing

Institutional repository: Faculty of Electrical Engineering, Computer Science and Information Technology Osijek

Cite this document

Vodička, D. (2023). Razdvajanje instrumentala i glasa u audio datoteci (Undergraduate thesis). Osijek: Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek. Retrieved from https://urn.nsk.hr/urn:nbn:hr:200:309870

Vodička, David. "Razdvajanje instrumentala i glasa u audio datoteci." Undergraduate thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, 2023. https://urn.nsk.hr/urn:nbn:hr:200:309870

Vodička, David. "Razdvajanje instrumentala i glasa u audio datoteci." Undergraduate thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, 2023. https://urn.nsk.hr/urn:nbn:hr:200:309870

Vodička, D. (2023). 'Razdvajanje instrumentala i glasa u audio datoteci', Undergraduate thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, accessed 11 May 2024, https://urn.nsk.hr/urn:nbn:hr:200:309870

Vodička D. Razdvajanje instrumentala i glasa u audio datoteci [Undergraduate thesis]. Osijek: Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek; 2023 [cited 2024 May 11] Available at: https://urn.nsk.hr/urn:nbn:hr:200:309870

D. Vodička, "Razdvajanje instrumentala i glasa u audio datoteci", Undergraduate thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, Osijek, 2023. Available at: https://urn.nsk.hr/urn:nbn:hr:200:309870

Metadata
TitleRazdvajanje instrumentala i glasa u audio datoteci
Title (english)Separation of instrumental and voice in an audio file
AuthorDavid Vodička
MentorHrvoje Leventić (mentor)
GranterJosip Juraj Strossmayer University of Osijek
Faculty of Electrical Engineering, Computer Science and Information Technology Osijek
(Department of Software Engineering)
(Chair of Visual Computing)
Osijek
Defense date and country2023, Croatia
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Computing
Data Processing
Abstract
Razdvajanje vokala i instrumenata je proces u kojemu uzimamo audio datoteku sa pomiješanim instrumentima i vokalom te pomoću raznih metoda i alata pokušavamo „izvući“ vokal od ostatka pjesme. Jednostavne metode koje i nisu tako učinkovite podrazumijevaju upotrebu Fourierove transformacije u kombinaciji sa filterima i maskama da bi dobili rezultat sa puno preklapanja instrumenata u kanalu koji bi trebao imati samo vokal. Sa druge strane imamo modele koji se temelje na neuronskim mrežama i učenju istih na velikom skupu podataka koji ima dostupno studijski snimljenu pjesmu i samo vokal pa je moguće jako dobro istrenirati model i dobiti rezultate visoke kvalitete. Modeli poput Demucs-a se jako dobro snalaze i u pjesmama sa vokalom koji nije toliko glasan ili je prikriven sa puno pozadinske buke, odnosno u pjesmama koje imaju izražen bas, bubnjeve ili ostale instrumente.
Abstract (english)
Separating vocals and instruments is a process in which we take an audio file with mixed instruments and vocals and try to "extract" the vocals from the rest of the song using various methods and tools. Simple methods that aren't that effective involve using a Fourier transform in combination with filters and masks to get a result with a lot of overlapping instruments in a channel that should only have vocals. On the other hand, we have models that are based on neural networks and learning them on a large data set that has a studio-recorded song and only vocals available, so it is possible to train the model very well and get high-quality results. Models like Demucs do very well in songs with vocals that are not that loud or are hidden with a lot of background noise, that is, in songs that have pronounced bass, drums or other instruments.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:200:309870
Study programmeTitle: Undergraduate University Study Programme in Computer Engineering
Study programme type: university
Study level: undergraduate
Academic / professional title: sveučilišni/a prvostupnik/prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
Type of resourceText
File originBorn digital
Access conditionsOpen access
Terms of useLicence In copyright icon
Created on2023-10-20 09:15:58