# LeVo 2: Stable and Melodious Song Generation via Hierarchical Representation Modeling and Progressive Post-Training

arXiv: https://arxiv.org/abs/2606.30642 · HTML: https://mlsanity.com/paper/2606.30642/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Builds on 48 of its 52 references that are in the map.
- First posted on arXiv 29 Jun 2026.

## Builds on (most cited first, 30 of 48)

- [High-Resolution Image Synthesis with Latent Diffusion Models](https://mlsanity.com/paper/2112.10752/) (2021)
- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [LLaMA: Open and Efficient Foundation Language Models](https://mlsanity.com/paper/2302.13971/) (2023)
- [Scalable Diffusion Models with Transformers](https://mlsanity.com/paper/2212.09748/) (2022)
- [Qwen2.5 Technical Report](https://mlsanity.com/paper/2412.15115/) (2024)
- [Neural Discrete Representation Learning](https://mlsanity.com/paper/1711.00937/) (2017)
- [LaMDA: Language Models for Dialog Applications](https://mlsanity.com/paper/2201.08239/) (2022)
- [SoundStream: An End-to-End Neural Audio Codec](https://mlsanity.com/paper/2107.03312/) (2021)
- [AudioLM: a Language Modeling Approach to Audio Generation](https://mlsanity.com/paper/2209.03143/) (2022)
- [Qwen2-Audio Technical Report](https://mlsanity.com/paper/2407.10759/) (2024)
- [Jukebox: A Generative Model for Music](https://mlsanity.com/paper/2005.00341/) (2020)
- [Simple and Controllable Music Generation](https://mlsanity.com/paper/2306.05284/) (2023)
- [MusicLM: Generating Music From Text](https://mlsanity.com/paper/2301.11325/) (2023)
- [High-Fidelity Audio Compression with Improved RVQGAN](https://mlsanity.com/paper/2306.06546/) (2023)
- [AudioLDM 2: Learning Holistic Audio Generation with Self-supervised Pretraining](https://mlsanity.com/paper/2308.05734/) (2023)
- [DiffSinger: Singing Voice Synthesis via Shallow Diffusion Mechanism](https://mlsanity.com/paper/2105.02446/) (2021)
- [MuseGAN: Multi-track Sequential Generative Adversarial Networks for Symbolic Music Generation and Accompaniment](https://mlsanity.com/paper/1709.06298/) (2017)
- [A Hierarchical Latent Vector Model for Learning Long-Term Structure in Music](https://mlsanity.com/paper/1803.05428/) (2018)
- [Tango 2: Aligning Diffusion-based Text-to-Audio Generations through Direct Preference Optimization](https://mlsanity.com/paper/2404.09956/) (2024)
- [Mousai: Text-to-Music Generation with Long-Context Latent Diffusion](https://mlsanity.com/paper/2301.11757/) (2023)
- [MidiNet: A Convolutional Generative Adversarial Network for Symbolic-domain Music Generation](https://mlsanity.com/paper/1703.10847/) (2017)
- [YuE: Scaling Open Foundation Models for Long-Form Music Generation](https://mlsanity.com/paper/2503.08638/) (2025)
- [MuQ: Self-Supervised Music Representation Learning with Mel Residual Vector Quantization](https://mlsanity.com/paper/2501.01108/) (2025)
- [ACE-Step: A Step Towards Music Generation Foundation Model](https://mlsanity.com/paper/2506.00045/) (2025)
- [Efficient Neural Music Generation](https://mlsanity.com/paper/2305.15719/) (2023)
- [Museformer: Transformer with Fine- and Coarse-Grained Attention for Music Generation](https://mlsanity.com/paper/2210.10349/) (2022)
- [DiffRhythm: Blazingly Fast and Embarrassingly Simple End-to-End Full-Length Song Generation with Latent Diffusion](https://mlsanity.com/paper/2503.01183/) (2025)
- [HiFiSinger: Towards High-Fidelity Neural Singing Voice Synthesis](https://mlsanity.com/paper/2009.01776/) (2020)
- [VISinger: Variational Inference with Adversarial Learning for End-to-End Singing Voice Synthesis](https://mlsanity.com/paper/2110.08813/) (2021)
- [MusicRL: Aligning Music Generation to Human Preferences](https://mlsanity.com/paper/2402.04229/) (2024)

## Built on by (most cited first, 6 of 6)

- [Pushing the Frontier of Full-Song Generation: Hierarchical Autoregressive Planning Meets Flow-Matching Rendering](https://mlsanity.com/paper/2607.20253/) (2026)
- [DiffSynth-Music: Audio-Conditioned KV-Cache Adapters for Controllable Music Generation](https://mlsanity.com/paper/2609.12774/) (2026)
- [AURA: Unified Multimodal Framework for Conversational Music Editing](https://mlsanity.com/paper/2609.14344/) (2026)
- [Open-Qwen-Music: An Auditable Framework for LLM-Based Music Composition and Diffusion Rendering](https://mlsanity.com/paper/2609.31652/) (2026)
- [YuE2: Unifying Symbolic and Audio Music Generation at Frontier Quality](https://mlsanity.com/paper/2609.33757/) (2026)
- [MIRA: A Musical Intent Refinement Agent for Aligning Text-to-Music Generation with User Intent](https://mlsanity.com/paper/2610.10355/) (2026)

## Abstract (arXiv)

Full-length song generation must preserve coherence and musicality, render detailed vocal and accompaniment acoustics, and follow lyrics and prompts. Existing language model-based systems face a structural trade-off: mixed-token modeling preserves vocal-instrument coordination but obscures track-specific details, whereas dual-track prediction improves acoustics but requires longer sequences and weakens global planning. We present LeVo 2, a hybrid LLM-Diffusion framework for controllable full-length song generation. LeVo 2 formulates this trade-off as hierarchical modeling: LeLM first predicts mixed tokens for semantic planning, then predicts vocal and accompaniment tokens in parallel for track-specific refinement, while a diffusion-based Music Codec reconstructs full-length waveforms. A central contribution of this extended version is an aesthetics-guided training schedule for alignment. During pre-training, an automated music aesthetic evaluation framework assigns musicality-tier conditions to large-scale data, providing musicality priors before preference alignment. Progressive post-training applies SFT, large-scale offline DPO, and closed-loop semi-online DPO to separately improve generation quality, controllability, and musicality. Modular extension then trains the Track-Specific LM for acoustic refinement while preserving the aligned semantic planner. This schedule separates musicality learning, controllability alignment, and acoustic refinement, mitigating optimization conflict and the limitations of static offline preference pairs. Expert listening tests and objective evaluations show that LeVo 2 outperforms open-source baselines across six subjective dimensions, and approaches leading commercial systems on several listening metrics. Ablations validate the effects of the training strategy, aesthetics guidance, scaling, and hierarchical architecture.