Discrete Diffusion Probabilistic Models for Symbolic Music Generation
This work addresses music generation for creative applications, but is incremental as it adapts an existing method to a new domain.
The authors tackled symbolic music generation by applying Discrete Diffusion Probabilistic Models (D3PMs) to polyphonic music, achieving state-of-the-art sample quality with flexible note-level infilling and post-hoc classifier guidance.
Denoising Diffusion Probabilistic Models (DDPMs) have made great strides in generating high-quality samples in both discrete and continuous domains. However, Discrete DDPMs (D3PMs) have yet to be applied to the domain of Symbolic Music. This work presents the direct generation of Polyphonic Symbolic Music using D3PMs. Our model exhibits state-of-the-art sample quality, according to current quantitative evaluation metrics, and allows for flexible infilling at the note level. We further show, that our models are accessible to post-hoc classifier guidance, widening the scope of possible applications. However, we also cast a critical view on quantitative evaluation of music sample quality via statistical metrics, and present a simple algorithm that can confound our metrics with completely spurious, non-musical samples.