A zero-training VLM framework generates music from images via ABC notation, multi-modal RAG, and self-refinement while providing text and visual explanations for the outputs.
On the evaluation of generative models in music,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SD 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Zero-Effort Image-to-Music Generation: An Interpretable RAG-based VLM Approach
A zero-training VLM framework generates music from images via ABC notation, multi-modal RAG, and self-refinement while providing text and visual explanations for the outputs.