Rethinking Music Captioning with Music Metadata LLMs

TL;DR

This paper introduces a metadata-based music captioning approach that predicts detailed music metadata from audio and uses LLMs to generate expressive captions, offering flexibility and efficiency over traditional end-to-end methods.

Contribution

It proposes a novel metadata prediction and captioning pipeline that reduces training time and allows style customization, improving over existing captioning models.

Findings

Comparable performance to end-to-end models with less training

Flexible caption stylization post-training

Effective metadata imputation and in-filling capabilities

Abstract

Music captioning, or the task of generating a natural language description of music, is useful for both music understanding and controllable music generation. Training captioning models, however, typically requires high-quality music caption data which is scarce compared to metadata (e.g., genre, mood, etc.). As a result, it is common to use large language models (LLMs) to synthesize captions from metadata to generate training data for captioning models, though this process imposes a fixed stylization and entangles factual information with natural language style. As a more direct approach, we propose metadata-based captioning. We train a metadata prediction model to infer detailed music metadata from audio and then convert it into expressive captions via pre-trained LLMs at inference time. Compared to a strong end-to-end baseline trained on LLM-generated captions derived from metadata, our method: (1) achieves comparable performance in less training time over end-to-end captioners, (2) offers flexibility to easily change stylization post-training, enabling output captions to be tailored to specific stylistic and quality requirements, and (3) can be prompted with audio and partial metadata to enable powerful metadata imputation or in-filling--a common task for organizing music data.