Original title: Voice and Music Separation Tool for Speech Transcription Support
Translated title: Voice and Music Separation Tool for Speech Transcription Support
Authors: Lashin, Mesk ; Schimmel, Jiří (referee) ; Frýza, Tomáš (advisor)
Document type: Bachelor's theses
Year: 2026
Language: eng
Publisher: Vysoké učení technické v Brně. Fakulta elektrotechniky a komunikačních technologií
Abstract: The goal of this thesis is to create and build a software application to separate musical recordings into vocals and instruments for the purpose of creating automatic speech-to-text transcripts. It focusses on the specific challenge that French rap presents to speech recognition systems i.e. high tempo, heavy use of instruments and loose language. the thesis reviews the various techniques presently in use for separating music tracks out of audio, the new machine learning approaches being developed, and then suggests a method to produce transcribable vocal track from the two previously mentioned methods. The thesis reviews and compares several separation algorithms and assesses them through the use of audio quality metrics and transcription-related performance metrics. The end product is a software application that shows how vocal separation will increase the expressiveness of the musical expression and result in an increase in the accuracy of creating transcripts for complex music.
Keywords: Audio Source Separation; Automatic Speech Recognition (ASR); Deep Learning for Audio; Music Signal Processing; Vocal–Instrumental Isolation; Audio Source Separation; Automatic Speech Recognition (ASR); Deep Learning for Audio; Music Signal Processing; Vocal–Instrumental Isolation

Institution: Brno University of Technology (web)
Document availability information: Fulltext is available in the Brno University of Technology Digital Library.
Original record: http://hdl.handle.net/11012/259255

Permalink: http://www.nusl.cz/ntk/nusl-776939


The record appears in these collections:
Universities and colleges > Public universities > Brno University of Technology
Academic theses (ETDs) > Bachelor's theses