MIRA: A Musical Intent Refinement Agent for Aligning Text-to-Music Generation with User Intent

Key points

• Text-to-music systems produce increasingly convincing audio, yet evaluation reveals little about whether the result matches user intent.1
• A global text-audio relevance score can overlook the implicit intent in underspecified prompts and mask failures in specific requirements, such as instrumentation, structure, rhythm, or mood progression.1
• We instantiate this as MuRA-Bench, a benchmark of real-world platform requests curated by music experts.1
• We further propose MIRA (Musical Intent Refinement Agent), a test-time agent that first grounds a request's intent into rubrics, then searches over prompt revisions for a black-box generator under a bounded budget, iteratively generating music, verifying it against the rubrics, and using this feedback to guide a trajectory-aware tree search.1

Sources (2)

• [1]MIRA: A Musical Intent Refinement Agent for Aligning Text-to-Music Generation with User Intent
  Hugging Face Daily Papers · Oct 7, 12:00 AM

  To bridge this gap, we formulate text-to-music intent alignment as satisfying a per-request rubric of independently verifiable items covering both a request's explicit requirements and its implied musical intent.

  Text-to-music systems produce increasingly convincing audio, yet evaluation reveals little about whether the result matches user intent.

• [2]MIRA: A Musical Intent Refinement Agent for Aligning Text-to-Music Generation with User Intent
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 7, 04:29 PM · same content

Extractive summary: sentences quoted from the sources.