Enhancing Singing Technique Control in SVS

Created using ChatSlide

The study introduces SinTechSVS, a system enhancing singing voice synthesis with precise singing technique controls. It highlights the evolution of SVS using deep learning, addresses challenges in emotional expressiveness, and proposes a Singing Technique Recommendation (STR) task. The methodology includes a Singing Technique Classifier and conditional synthesis integration. Experiments use an expert-annotated dataset to evaluate synthesis quality, showcasing STR effectiveness. Results...