Performance Conditioning for Diffusion-Based Multi-Instrument Music Synthesis

TL;DR

This paper introduces a performance conditioning method for diffusion-based multi-instrument music synthesis, enabling control over style and timbre based on specific performances, leading to more realistic and customizable generated music.

Contribution

It proposes a novel performance conditioning technique for diffusion models, improving control over style and timbre in multi-instrument music synthesis.

Findings

Achieved state-of-the-art FAD realism scores.

Enabled novel timbre and style control.

Demonstrated effectiveness on diverse uncurated performances.

Abstract

Generating multi-instrument music from symbolic music representations is an important task in Music Information Retrieval (MIR). A central but still largely unsolved problem in this context is musically and acoustically informed control in the generation process. As the main contribution of this work, we propose enhancing control of multi-instrument synthesis by conditioning a generative model on a specific performance and recording environment, thus allowing for better guidance of timbre and style. Building on state-of-the-art diffusion-based music generative models, we introduce performance conditioning - a simple tool indicating the generative model to synthesize music with style and timbre of specific instruments taken from specific performances. Our prototype is evaluated using uncurated performances with diverse instrumentation and achieves state-of-the-art FAD realism scores while allowing novel timbre and style control. Our project page, including samples and demonstrations, is available at benadar293.github.io/midipm