Symbolic Music Generation Conditioned on Continuous-Valued Emotions

In this paper we present a new approach for the generation of multi-instrument symbolic music driven by musical emotion. The principal novelty our centres on conditioning state-of-the-art transformer based continuous-valued valence and arousal labels. addition, provide large-scale dataset paired with emotion labels in terms arousal. We evaluate quantitative manner two ways, first measuring its note prediction accuracy, second via regression task valence-arousal plane. Our results demonstrate that proposed approaches outperform using control tokens which is representative current state art.