MusicDiffusionNet: Enhancing Text-to-Music Generation with Adaptive Style and Multi-Scale Temporal Mixup Strategies | Synapse