Abstract
Existing symbolic music generation models typically use bars as the basic structural unit. However, human perception of musical phrases often does not align with notated bar lines, leading to long-term structural fragmentation. This paper proposes RPPNet—a two-stage deep learning architecture with variable structural boundaries.
Method
- Generating variable-length Rhythm-Pitch Primitive (RPP) sequences: Each RPP encodes note count, rhythm, and contour.
- Decoding RPP sequences into concrete notes: The grouping of RPPs is automatically derived from acoustic cues, auditory inertia, and similarity perception based on music psychology.
Results
Experiments show that melodies generated by RPPNet are superior in both long-term structure and musicality, with significant improvements across all subjective evaluation dimensions. Ablation studies confirm that the performance gain stems from the structural correctness of the psychological representation, rather than from model capacity.
Conclusion
This work offers an interdisciplinary perspective for music generation, integrating music theory, computational modeling, and music psychology.
Blogger's Review: The innovation of RPPNet lies in its perceptual grouping approach, enhancing the structural integrity of music generation and overcoming traditional models' limitations regarding bar divisions. Its methodology is worth referencing and applying in other fields.