SinTechSVS: Controllable Singing Voice Synthesis

SinTechSVS introduces enhanced techniques for singing voice synthesis (SVS), addressing limitations in expressiveness and aiming to control singing techniques. Key contributions include three innovative modules integrated with the DiffSinger framework and a focus on controllability through rigorous evaluation metrics. A detailed methodology covers system architecture, classifiers, and objective functions. Experiments utilise diverse datasets, investigating synthesis performance and...