# SketchSong: Hierarchical Song Generation with Sketch Planning and Fine-Grained Multi-Track Modeling

arXiv: https://arxiv.org/abs/2606.03169 · HTML: https://mlsanity.com/paper/2606.03169/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 20 of its 30 references that are in the map.
- First posted on arXiv 2 Jun 2026.

## Builds on (most cited first, 20 of 20)

- [Robust Speech Recognition via Large-Scale Weak Supervision](https://mlsanity.com/paper/2212.04356/) (2022)
- [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://mlsanity.com/paper/2006.11477/) (2020)
- [Jukebox: A Generative Model for Music](https://mlsanity.com/paper/2005.00341/) (2020)
- [MusicLM: Generating Music From Text](https://mlsanity.com/paper/2301.11325/) (2023)
- [Fr'echet Audio Distance: A Metric for Evaluating Music Enhancement Algorithms](https://mlsanity.com/paper/1812.08466/) (2018)
- [Meta Audiobox Aesthetics: Unified Automatic Quality Assessment for Speech, Music, and Sound](https://mlsanity.com/paper/2502.05139/) (2025)
- [YuE: Scaling Open Foundation Models for Long-Form Music Generation](https://mlsanity.com/paper/2503.08638/) (2025)
- [All-In-One Metrical And Functional Structure Analysis With Neighborhood Attentions on Demixed Audio](https://mlsanity.com/paper/2307.16425/) (2023)
- [Joint Audio and Symbolic Conditioning for Temporally Controlled Text-to-Music Generation](https://mlsanity.com/paper/2406.10970/) (2024)
- [SongEval: A Benchmark Dataset for Song Aesthetics Evaluation](https://mlsanity.com/paper/2505.10793/) (2025)
- [ACE-Step 1.5: Pushing the Boundaries of Open-Source Music Generation](https://mlsanity.com/paper/2602.00744/) (2026)
- [SongBloom: Coherent Song Generation via Interleaved Autoregressive Sketching and Diffusion Refinement](https://mlsanity.com/paper/2506.07634/) (2025)
- [Seed-Music: A Unified Framework for High Quality and Controlled Music Generation](https://mlsanity.com/paper/2409.09214/) (2024)
- [JEN-1 Composer: A Unified Framework for High-Fidelity Multi-Track Music Generation](https://mlsanity.com/paper/2310.19180/) (2023)
- [MusicFlow: Cascaded Flow Matching for Text Guided Music Generation](https://mlsanity.com/paper/2410.20478/) (2024)
- [DiffRhythm 2: Efficient and High Fidelity Song Generation via Block Flow Matching](https://mlsanity.com/paper/2510.22950/) (2025)
- [Analyzable Chain-of-Musical-Thought Prompting for High-Fidelity Music Generation](https://mlsanity.com/paper/2503.19611/) (2025)
- [MusiConGen: Rhythm and Chord Control for Transformer-Based Text-to-Music Generation](https://mlsanity.com/paper/2407.15060/) (2024)
- [Muse: Towards Reproducible Long-Form Song Generation with Fine-Grained Style Control](https://mlsanity.com/paper/2601.03973/) (2026)
- [SegTune: Structured and Fine-Grained Control for Song Generation](https://mlsanity.com/paper/2510.18416/) (2025)

## Abstract (arXiv)

Recent song generation systems can synthesize realistic audio, yet generating complete songs remains challenging for two reasons. First, explicit song-level arrangement planning remains limited in existing methods, so models often need to organize overall arrangement development while generating low-level audio details. This often leads to incoherence in arrangements, such as weak section transitions and limited dynamic progression. Second, coarse modeling of different musical parts obscures their distinct roles and interactions, limiting arrangement richness of generated songs. In this paper, we present SketchSong, a hierarchical song generation framework that addresses these issues through song-level sketch planning and fine-grained multi-track modeling. Along the temporal dimension, SketchSong first predicts a compact sequence of high-level sketch tokens derived from compressed audio representations, and then generates audio tokens conditioned on these sketches. This coarse-to-fine process gives the model an explicit arrangement plan before detailed audio generation. Along the track dimension, SketchSong explicitly models four tracks, i.e., vocals, bass, drums and other instruments. This enables the model to capture the roles and interactions of different musical parts more precisely. Experiments on song generation benchmarks show that SketchSong consistently outperforms our baseline on both objective metrics and human listening tests. Despite not employing additional post-training for preference optimization such as lyrics and text-prompt alignments, SketchSong achieves competitive results against strong, post-trained open-source systems, demonstrating the effectiveness of our overall design.