Instruct-MusicGen: Unlocking Text-to-Music Editing for Music Language Models via Instruction Tuning

2024-05-28

arXiv.org

AI Summary

Text-to-music editing faces two key bottlenecks: task-specific models require costly from-scratch training, while LLM-based approaches suffer from severe audio reconstruction distortion. This paper introduces the first instruction-tuning paradigm tailored for music language models—enabling efficient adaptation to editing tasks without full model retraining. Our method adds only 8% parameters and requires just 5K training steps. Built upon the MusicGen architecture, it incorporates dual fusion modules—text fusion and audio fusion—that jointly encode instruction text and the original audio waveform. The framework supports fine-grained edits, including instrument addition/removal and track separation. Experiments demonstrate substantial improvements: +12.3% instruction-following accuracy and −18.7% Mel-Cepstral Distortion (MCD), indicating superior audio fidelity. Our approach consistently outperforms all baselines across diverse music editing tasks.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Generation
• Computer Vision: Diffusion Models for Vision

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Recent advances in text-to-music editing, which employ text queries to modify music (e.g. by changing its style or adjusting instrumental components), present unique challenges and opportunities for AI-assisted music creation. Previous approaches in this domain have been constrained by the necessity to train specific editing models from scratch, which is both resource-intensive and inefficient; other research uses large language models to predict edited music, resulting in imprecise audio reconstruction. To Combine the strengths and address these limitations, we introduce Instruct-MusicGen, a novel approach that finetunes a pretrained MusicGen model to efficiently follow editing instructions such as adding, removing, or separating stems. Our approach involves a modification of the original MusicGen architecture by incorporating a text fusion module and an audio fusion module, which allow the model to process instruction texts and audio inputs concurrently and yield the desired edited music. Remarkably, Instruct-MusicGen only introduces 8% new parameters to the original MusicGen model and only trains for 5K steps, yet it achieves superior performance across all tasks compared to existing baselines, and demonstrates performance comparable to the models trained for specific tasks. This advancement not only enhances the efficiency of text-to-music editing but also broadens the applicability of music language models in dynamic music production environments.

Problem

• Enables text-based music editing via instruction tuning
• Overcomes inefficiency of training separate editing models
• Improves precision in audio reconstruction for edits

Innovation

• Finetunes pretrained MusicGen for editing instructions
• Incorporates text and audio fusion modules
• Adds only 8% new parameters for efficiency