ACE-Step 1.5
Pushing the Boundaries of Open-Source Music Generation
Project | Hugging Face | ModelScope | Space Demo | Discord Tech Report
Model Details
๐ ACE-Step v1.5 is a highly efficient open-source music foundation model designed to bring commercial-grade music generation to consumer hardware.
Key Features
- ๐ฐ Commercial-Ready: Unlike many models trained on ambiguous datasets, ACE-Step v1.5 is designed for creators. You can strictly use the generated music for commercial purposes.
- ๐ Safe & Robust Training Data: The model is trained on a massive, legally compliant dataset consisting of:
- Licensed Data: Professionally licensed music tracks.
- Royalty-Free / No-Copyright Data: A vast collection of public domain and royalty-free music.
- Synthetic Data: High-quality audio generated via advanced MIDI-to-Audio conversion.
- โก Extreme Speed: Generates a full song in under 2 seconds on an A100 and under 10 seconds on an RTX 3090.
- ๐ฅ๏ธ Consumer Hardware Friendly: Runs locally with less than 4GB of VRAM.
Technical Capabilities
๐ At its core lies a novel hybrid architecture where the Language Model (LM) functions as an omni-capable planner: it transforms simple user queries into comprehensive song blueprintsโscaling from short loops to 10-minute compositionsโwhile synthesizing metadata, lyrics, and captions via Chain-of-Thought to guide the Diffusion Transformer (DiT). โก Uniquely, this alignment is achieved through intrinsic reinforcement learning relying solely on the model's internal mechanisms, thereby eliminating the biases inherent in external reward models or human preferences. ๐๏ธ
๐ฎ Beyond standard synthesis, ACE-Step v1.5 unifies precise stylistic control with versatile editing capabilitiesโsuch as cover generation, repainting, and vocal-to-BGM conversionโwhile maintaining strict adherence to prompts across 50+ languages. This paves the way for powerful tools that seamlessly integrate into the creative workflows of music artists, producers, and content creators. ๐ธ
- Developed by: [ACE-STEP]
- Model type: [Text2Music]
- Language(s): [50+ languages]
- License: [MIT]
Evaluation
๐๏ธ Architecture
๐ฆ Model Zoo
DiT Models
| DiT Model | Pre-Training | SFT | RL | CFG | Step | Refer audio | Text2Music | Cover | Repaint | Extract | Lego | Complete | Quality | Diversity | Fine-Tunability | Hugging Face |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
acestep-v15-base |
โ | โ | โ | โ | 50 | โ | โ | โ | โ | โ | โ | โ | Medium | High | Easy | Link |
acestep-v15-sft |
โ | โ | โ | โ | 50 | โ | โ | โ | โ | โ | โ | โ | High | Medium | Easy | Link |
acestep-v15-turbo |
โ | โ | โ | โ | 8 | โ | โ | โ | โ | โ | โ | โ | Very High | Medium | Medium | Link |
acestep-v15-turbo-rl |
โ | โ | โ | โ | 8 | โ | โ | โ | โ | โ | โ | โ | Very High | Medium | Medium | To be released |
LM Models
| LM Model | Pretrain from | Pre-Training | SFT | RL | CoT metas | Query rewrite | Audio Understanding | Composition Capability | Copy Melody | Hugging Face |
|---|---|---|---|---|---|---|---|---|---|---|
acestep-5Hz-lm-0.6B |
Qwen3-0.6B | โ | โ | โ | โ | โ | Medium | Medium | Weak | โ |
acestep-5Hz-lm-1.7B |
Qwen3-1.7B | โ | โ | โ | โ | โ | Medium | Medium | Medium | โ |
acestep-5Hz-lm-4B |
Qwen3-4B | โ | โ | โ | โ | โ | Strong | Strong | Strong | โ |
๐ Acknowledgements
This project is co-led by ACE Studio and StepFun.
๐ Citation
If you find this project useful for your research, please consider citing:
@misc{gong2026acestep,
title={ACE-Step 1.5: Pushing the Boundaries of Open-Source Music Generation},
author={Junmin Gong, Yulin Song, Wenxiao Zhao, Sen Wang, Shengyuan Xu, Jing Guo},
howpublished={\url{https://github.com/ace-step/ACE-Step-1.5}},
year={2026},
note={GitHub repository}
}
- Downloads last month
- 75



