Music Style Transfer With Diffusion Model

TL;DR

The paper tackles the challenge of many-to-many music style transfer by introducing a diffusion-model framework that operates on spectrograms within a perceptually compressed latent space. Style transfer is achieved via a latent diffusion model with cross-attention conditioning, while GuideDiff serves as a fast, high-fidelity waveform decoder. The approach yields real-time performance on consumer GPUs and strong results in instrument timbre and compositional style transfer, with competitive audio quality compared to autoregressive vocoders. This work provides a scalable, practical path for flexible music style migration using diffusion-driven spectrograms and diffusion-guided waveform reconstruction.

Abstract

Previous studies on music style transfer have mainly focused on one-to-one style conversion, which is relatively limited. When considering the conversion between multiple styles, previous methods required designing multiple modes to disentangle the complex style of the music, resulting in large computational costs and slow audio generation. The existing music style transfer methods generate spectrograms with artifacts, leading to significant noise in the generated audio. To address these issues, this study proposes a music style transfer framework based on diffusion models (DM) and uses spectrogram-based methods to achieve multi-to-multi music style transfer. The GuideDiff method is used to restore spectrograms to high-fidelity audio, accelerating audio generation speed and reducing noise in the generated audio. Experimental results show that our model has good performance in multi-mode music style transfer compared to the baseline and can generate high-quality audio in real-time on consumer-grade GPUs.

Figures (6)

• Figure 1: Piano to violin style transfer.
• Figure 2: Spectrogram.
• Figure 3: Models of transfer.
• Figure 4: GuideDiff architecture.
• Figure 5: Model's primary network.