# Text2midi-InferAlign: Improving Symbolic Music Generation with Inference-Time Alignment

arXiv: https://arxiv.org/abs/2505.12669 · HTML: https://mlsanity.com/paper/2505.12669/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Builds on 21 of its 35 references that are in the map.
- First posted on arXiv 19 May 2025.

## Builds on (most cited first, 21 of 21)

- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [Scaling Instruction-Finetuned Language Models](https://mlsanity.com/paper/2210.11416/) (2022)
- [Large-scale Contrastive Language-Audio Pretraining with Feature Fusion and Keyword-to-Caption Augmentation](https://mlsanity.com/paper/2211.06687/) (2022)
- [Simple and Controllable Music Generation](https://mlsanity.com/paper/2306.05284/) (2023)
- [Noise2Music: Text-conditioned Music Generation with Diffusion Models](https://mlsanity.com/paper/2302.03917/) (2023)
- [ARGS: Alignment as Reward-Guided Search](https://mlsanity.com/paper/2402.01694/) (2024)
- [Mustango: Toward Controllable Text-to-Music Generation](https://mlsanity.com/paper/2311.08355/) (2023)
- [DeepBach: a Steerable Model for Bach Chorales Generation](https://mlsanity.com/paper/1612.01010/) (2016)
- [BOND: Aligning LLMs with Best-of-N Distillation](https://mlsanity.com/paper/2407.14622/) (2024)
- [LLM-based Multi-Agent Reinforcement Learning: Current and Future Directions](https://mlsanity.com/paper/2405.11106/) (2024)
- [A Comprehensive Review of Multimodal Large Language Models: Performance and Challenges Across Different Tasks](https://mlsanity.com/paper/2408.01319/) (2024)
- [A Comprehensive Survey on Deep Music Generation: Multi-level Representations, Algorithms, Evaluations, and Future Directions](https://mlsanity.com/paper/2011.06801/) (2020)
- [MuseCoco: Generating Symbolic Music from Text](https://mlsanity.com/paper/2306.00110/) (2023)
- [Museformer: Transformer with Fine- and Coarse-Grained Attention for Music Generation](https://mlsanity.com/paper/2210.10349/) (2022)
- [MusicRL: Aligning Music Generation to Human Preferences](https://mlsanity.com/paper/2402.04229/) (2024)
- [MidiCaps: A large-scale MIDI dataset with text captions](https://mlsanity.com/paper/2406.02255/) (2024)
- [Text2midi: Generating Symbolic Music from Captions](https://mlsanity.com/paper/2412.16526/) (2024)
- [Hierarchical Recurrent Neural Networks for Conditional Melody Generation with Long-term Structure](https://mlsanity.com/paper/2102.09794/) (2021)
- [MIDI-GPT: A Controllable Generative Model for Computer-Assisted Multitrack Music Composition](https://mlsanity.com/paper/2501.17011/) (2025)
- [Midi Miner -- A Python library for tonal tension and track classification](https://mlsanity.com/paper/1910.02049/) (2019)
- [Inference Time Alignment with Reward-Guided Tree Search](https://mlsanity.com/paper/2406.15193/) (2024)

## Built on by (most cited first, 4 of 4)

- [MelodySim: Measuring Melody-aware Music Similarity for Plagiarism Detection](https://mlsanity.com/paper/2505.20979/) (2025)
- [Amadeus: Autoregressive Model with Bidirectional Attribute Modelling for Symbolic Music](https://mlsanity.com/paper/2508.20665/) (2025)
- [Aligning Generative Music AI with Human Preferences: Methods and Challenges](https://mlsanity.com/paper/2511.15038/) (2025)
- [MIRA: A Musical Intent Refinement Agent for Aligning Text-to-Music Generation with User Intent](https://mlsanity.com/paper/2610.10355/) (2026)

## Abstract (arXiv)

We present Text2midi-InferAlign, a novel technique for improving symbolic music generation at inference time. Our method leverages text-to-audio alignment and music structural alignment rewards during inference to encourage the generated music to be consistent with the input caption. Specifically, we introduce two objectives scores: a text-audio consistency score that measures rhythmic alignment between the generated music and the original text caption, and a harmonic consistency score that penalizes generated music containing notes inconsistent with the key. By optimizing these alignment-based objectives during the generation process, our model produces symbolic music that is more closely tied to the input captions, thereby improving the overall quality and coherence of the generated compositions. Our approach can extend any existing autoregressive model without requiring further training or fine-tuning. We evaluate our work on top of Text2midi - an existing text-to-midi generation model, demonstrating significant improvements in both objective and subjective evaluation metrics.