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

Zekai Liu; Zhilin Wang; Xuzheng He; Yu Cheng; Yang Yang

cs.SDOct 7, 2026

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

Authors: Zekai Liu, Zhilin Wang, Xuzheng He, Yu Cheng, Yang Yang

Organizations: Shandong University · University of Science and Technology of China · Central Conservatory of Music · Kunlun Tech Co. Ltd. · Shanghai Jiao Tong University

Abstract

Text-to-music systems produce increasingly convincing audio, yet evaluation reveals little about whether the result matches user intent. 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. 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. Scoring items individually makes evaluation diagnostic by intent source and musical dimension, rather than a single opaque score. We instantiate this as MuRA-Bench, a benchmark of real-world platform requests curated by music experts. 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. Experiments across open-source and commercial backends show that MIRA improves intent alignment, enabling an open-source generator to achieve performance comparable to representative commercial systems (e.g. Suno and Mureka). Project page: https://mirareview.github.io/.

Figures & tables

Figure 1: Overview of MuRA-Bench construction and evaluation. Expert-revised, request-specific rubrics support fine-grained assessment with MusicFlamingo. Gold rubrics are used exclusively for evaluation and are not exposed to the generation system.

Figure 2: Overview of MIRA and its verifier-guided refinement process.

	Item-level		Clip-level			Pair
Evaluator	SRCC	$\tau_{b}$	LCC	SRCC	$\tau_{b}$	Acc. (%)
MusicFlamingo	.690	.532	.817	.823	.646	79.59
Qwen3-Omni-30B-A3B	.509	.375	.632	.631	.480	74.15
Qwen2.5-Omni-7B	.469	.344	.599	.643	.465	72.79
MOSS-Music-8B	.502	.371	.591	.593	.429	62.59
AnyAudio-Judge-7B	.462	.335	.561	.556	.393	68.03

Table 1: Agreement with expert ratings on 100 clips from 25 requests. The first five evaluators use AQA. MF denotes MusicFlamingo.

	Intent alignment				Musical dimensions
Backend / Setting	Overall	D-Macro	Base	Comp.	Style	I/V	Mood	Rhythm	H/M	S/E	P/T
ACE-Step v1.5 Turbo
Direct	67.6	67.2	69.6	66.1	68.1	62.4	71.9	71.0	64.1	69.4	63.3
MIRA ( $B=3$ )	74.2 $\uparrow$ 6.6	74.1 $\uparrow$ 6.9	77.1 $\uparrow$ 7.5	71.1 $\uparrow$ 5.0	74.6 $\uparrow$ 6.5	69.5 $\uparrow$ 7.1	81.2 $\uparrow$ 9.3	75.4 $\uparrow$ 4.4	71.4 $\uparrow$ 7.3	72.8 $\uparrow$ 3.4	73.7 $\uparrow$ 10.4
MIRA ( $B=9$ )	76.3 $\uparrow$ 8.7	75.6 $\uparrow$ 8.4	79.1 $\uparrow$ 9.5	73.4 $\uparrow$ 7.3	75.4 $\uparrow$ 7.3	69.1 $\uparrow$ 6.7	83.0 $\uparrow$ 11.1	77.2 $\uparrow$ 6.2	72.8 $\uparrow$ 8.7	74.6 $\uparrow$ 5.2	77.0 $\uparrow$ 13.7
SongGeneration2 Large

Table 2: Results on the full MuRA-Bench benchmark. All scores are multiplied by 100; higher is better. Superscripts in MIRA rows indicate absolute changes from the same backend’s direct-prompt baseline on this scale. Bold marks the best score within each refinement backend or among the commercial models.

Figure 3: Blinded expert ratings of overall, explicit, and inferred intent on MuRA-Bench. Bars show mean ratings on a 1–5 scale, with 95% confidence intervals.

		ACE-Step v1.5 Turbo				YuE2-3B
Setting	$B$	Overall	D-Macro	Base	Comp.	Overall	D-Macro	Base	Comp.
Direct	1	0.759	0.756	0.803	0.710	0.816	0.811	0.868	0.752
Best-of-9	9	0.800	0.795	0.874	0.719	0.848	0.850	0.910	0.775
Tools + Best-of-9	9	0.815	0.807	0.871	0.748	0.860	0.852	0.911	0.800
Tools + Adaptive search	9	0.826	0.814	0.862	0.781	0.867	0.863	0.913	0.808
Full MIRA	9	0.836	0.826	0.880	0.782	0.872	0.868	0.911	0.821

Table 3: Component analysis on a fixed subset of 20 MuRA-Bench requests. Direct generation uses

B=1

; all remaining configurations use

B=9

. Tool grounding is first combined with best-of-nine sampling, followed by adaptive search without memory and then tree-structured memory. Bold marks the best score within each backend.

Figure 4: Best-of-9 versus MIRA with

B=9

. Scores are multiplied by 100; the vertical axis starts at 75.

Appendix figures & tables5 assets

Supplementary material from the paper’s appendix.

Appendix

Rubric dimension	Items
Style	135
Instrumentation/vocal	303
Mood	218
Rhythm	257
Harmony/melody	114
Structure/energy	199

Appendix

Table 4: Distribution of the 1,566 rubric items across seven musical dimensions.

Source	Requirement	Dimension
Base	upbeat combat music for a 2D role-playing game	Style
Base	open-plains biome atmosphere	Mood
Base	flute parts	Instrumentation/vocal
Base	wind textures	Production/texture
Base	African drum rhythms	Rhythm
Base	energizing battle-ready mood	Mood

Appendix

Table 5: A complete expert-revised rubric for RPG battle music, comprising six base and six completion requirements.

20-point	5-point	Interpretation
1–4	1	Not fulfilled: the required property is absent or clearly contradicted.
5–8	2	Slightly fulfilled: only weak evidence is audible, with most of the requirement unmet.
9–12	3	Partially fulfilled: recognizable evidence is present, but substantial shortcomings remain.
13–16	4	Mostly fulfilled: the main requirement is realized, with minor shortcomings.
17–20	5	Fully fulfilled: the requirement is clearly and sufficiently realized, without an evident deviation.

Appendix

Table 6: Scoring anchors for item-level and system-level expert assessments.

Setting	Value or rule
Generation budget $B$	3 or 9, including the initial candidate
Low-score feedback threshold $\eta$	0.60
Memory change tolerance $\epsilon$	0.03
Online ranking score	Mean item satisfaction over the fixed online rubric
Complementary frontier criterion	Mean of the lowest-scoring quarter of items
Final selection	Highest-scoring successfully visited candidate

Appendix

Table 7: MIRA search settings and generation-budget conventions.

	Intent alignment				Musical dimensions
Backend / Setting	Overall	D-Macro	Base	Comp.	Style	I/V	Mood	Rhythm	H/M	S/E	P/T
Evaluator: MOSS-Music
ACE-Step v1.5 Turbo
Direct	64.7	65.6	66.5	63.8	69.0	55.8	68.9	67.5	69.1	69.7	59.0
MIRA ( $B=3$ )	71.2 $\uparrow$ 6.5	71.5 $\uparrow$ 5.9	73.1 $\uparrow$ 6.7	68.1 $\uparrow$ 4.3	73.3 $\uparrow$ 4.3	66.0 $\uparrow$ 10.2	80.1 $\uparrow$ 11.2	70.5 $\uparrow$ 3.0	70.9 $\uparrow$ 1.8	70.6 $\uparrow$ 0.9	68.8 $\uparrow$ 9.8
MIRA ( $B=9$ )	71.9 $\uparrow$ 7.1	72.7 $\uparrow$ 7.1	73.1 $\uparrow$ 6.6	69.8 $\uparrow$ 6.0	74.5 $\uparrow$ 5.5	59.9 $\uparrow$ 4.1	80.4 $\uparrow$ 11.4	72.2 $\uparrow$ 4.7	74.8 $\uparrow$ 5.7	73.8 $\uparrow$ 4.2	73.2 $\uparrow$ 14.2

Appendix

Table 8: Cross-evaluator assessment on MuRA-Bench. MOSS-Music and Qwen3-Omni score outputs from direct generation and MusicFlamingo-guided MIRA. Commercial direct-prompting baselines are evaluated with MOSS-Music and Qwen3-Omni. Scores are multiplied by 100; higher is better. Superscripts indicate differences from the reported Direct score within each backend–evaluator group. Bold marks the best reported score within each refinement group or among the commercial baselines, before rounding.

Explore similar work

TuneJury: An Open Metric for Improving Music Generation Preference Alignment
Jun 15, 2026Yonghyun Kim, Junwon Lee, Haiwen Xia +5Text-To-Music Song Generation
Georgia Tech · KAIST · Peking University +3
MusicLayout: Explicit Structural Planning for Controllable Text-to-Music Generation
Aug 10, 2026Shuyu Li, Kejun Zhang, Jiahe Lei +5Text-To-Music
College of Artificial Intelligence, Zhejiang University · 3Innovation Center of Yangtze River Delta, Zhejiang University · 4The Chinese University of Hong Kong +4
Rubric-Based Optimization for Text-to-Music Generation
Oct 2, 2026Ping Wang, Guang Yang, Shao-Rong Su +3Text-To-Music
Paul G. Allen School of Computer Science & Engineering, University of Washington · Department of Electrical and Computer Engineering, University of Washington · Allen Institute for AI

Jun 15, 2026, cs.SDJ/K move · Enter open · S save

TuneJury: An Open Metric for Improving Music Generation Preference Alignment

Yonghyun Kim, Junwon Lee, Haiwen Xia, Yinghao Ma+4

Georgia Tech · KAIST · Peking University+3

Text-To-Music Song Generation Preference Datasets Tuning

Open paper AskPDF arXiv

We introduce TuneJury, an open, instance-level pairwise reward model for text-to-music that predicts a music preference score from a text prompt and an audio clip. The released checkpoint is trained on publicly available human-preference labels covering arena-style (A vs. B) votes, metric-alignment preference pairs, crowdsourced pairwise comparisons, and expert aesthetic ratings. The predicted score margin between two clips is well calibrated on our held-out test split, supporting data filtering via a simple score threshold. TuneJury generalizes to both held-out test pairs and out-of-distribution benchmarks, remaining competitive with prior baselines on the latter. For generators released after training, we introduce anchor calibration, a post-hoc, per-system Bradley-Terry calibration that recovers agreement at substantially better data efficiency than from-scratch retraining. The same frozen reward drives consistent reward-axis gains across three downstream applications: inference-time best-of-N selection, DITTO-style latent optimization, and expert-iteration post-training. TuneJury is available at https://github.com/yonghyunk1m/TuneJury.

Abstract

Explore similar work

TuneJury: An Open Metric for Improving Music Generation Preference Alignment

MusicLayout: Explicit Structural Planning for Controllable Text-to-Music Generation

Rubric-Based Optimization for Text-to-Music Generation