https://github.com/tencent-ailab/SongBloom
The official code repository for SongBloom: Coherent Song Generation via Interleaved Autoregressive Sketching and Diffusion Refinement
https://github.com/tencent-ailab/SongBloom
Last synced: 5 months ago
JSON representation
The official code repository for SongBloom: Coherent Song Generation via Interleaved Autoregressive Sketching and Diffusion Refinement
- Host: GitHub
- URL: https://github.com/tencent-ailab/SongBloom
- Owner: tencent-ailab
- License: other
- Created: 2025-06-19T15:52:02.000Z (11 months ago)
- Default Branch: master
- Last Pushed: 2025-12-04T07:10:13.000Z (5 months ago)
- Last Synced: 2025-12-07T14:19:48.460Z (5 months ago)
- Language: Python
- Homepage:
- Size: 24.1 MB
- Stars: 682
- Watchers: 20
- Forks: 72
- Open Issues: 16
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
- StarryDivineSky - tencent-ailab/SongBloom - 精炼”的交替流程有效解决了传统音乐生成模型在结构完整性与细节丰富性之间的平衡难题。项目基于大规模音乐数据集训练，能够捕捉不同风格的音乐特征，生成的作品既符合音乐理论逻辑，又具备细腻的情感表达。SongBloom的官方代码已开源，为研究者和开发者提供了完整的实现方案，其核心贡献包括提出新型生成框架、设计双阶段协同机制以及验证了结合自回归与扩散模型在音乐生成领域的有效性，为高质量音乐创作提供了新的技术路径。 (语音合成 / 资源传输下载)
README
# **SongBloom**: *Coherent Song Generation via Interleaved Autoregressive Sketching and Diffusion Refinement*
[](https://arxiv.org/abs/2506.07634)
[](https://huggingface.co/CypressYang/SongBloom)
[](https://cypress-yang.github.io/SongBloom_demo)
We propose **SongBloom**, a novel framework for full-length song generation that leverages an interleaved paradigm of autoregressive sketching and diffusion-based refinement. SongBloom employs an autoregressive diffusion model that combines the high fidelity of diffusion models with the scalability of language models.
Specifically, it gradually extends a musical sketch from short to long and refines the details from coarse to fine-grained. The interleaved generation paradigm effectively integrates prior semantic and acoustic context to guide the generation process.
Experimental results demonstrate that SongBloom outperforms existing methods across both subjective and objective metrics and achieves performance comparable to the state-of-the-art commercial music generation platforms.
## Models
| Name | Size | Max Length | Prompt type | 🤗 |
| -------------------- | ---- | ---------- | ----------- | -------------------------------------------- |
| songbloom_full_150s | 2B | 2m30s | 10s wav | [link](https://huggingface.co/CypressYang/SongBloom) |
| songbloom_full_150s_dpo | 2B | 2m30s | 10s wav | [link](https://huggingface.co/CypressYang/SongBloom) |
| songbloom_full_240s$^{[1]}$ | 2B | 4m | 10s wav | [link](https://huggingface.co/CypressYang/SongBloom_long) |
| ... | | | | |
- [1] For the **_150s** series models, each `[intro]`, `[outro]`, and `[inst]` corresponds to an expected duration of 1 second; whereas for the **_240s** series models, each token corresponds to 5 seconds (details in [docs/lyric_format](docs/lyric_format.md)).
## Updates
- **Oct 2025**: Release songbloom_full_240s; fix bugs in half-precision inference ; Reduce GPU memory consumption during the VAE stage.
- **Sep 2025**: Release the songbloom_full_150s model with DPO post-training
- **Jun 2025**: Release the songbloom_full_150s and inference script
## Getting Started
### Prepare Environments
```bash
conda create -n SongBloom python==3.8.12
conda activate SongBloom
# yum install libsndfile
# pip install torch==2.2.0 torchaudio==2.2.0 --index-url https://download.pytorch.org/whl/cu118 # For different CUDA version
pip install -r requirements.txt
```
### Data Preparation
A .jsonl file, where each line is a json object:
```json
{
"idx": "The index of each sample",
"lyrics": "The lyrics to be generated",
"prompt_wav": "The path of the style prompt audio",
}
```
One example can be refered to as: [example/test.jsonl](example/test.jsonl)
The prompt wav should be a 10-second, 48kHz audio clip.
For details on lyric formatting, see [docs/lyric_format.md](docs/lyric_format.md).
### Inference
```bash
source set_env.sh
python3 infer.py --input-jsonl example/test.jsonl
# For GPUs with low VRAM like RTX4090, you should set the dtype as bfloat16
python3 infer.py --input-jsonl example/test.jsonl --dtype bfloat16
# SongBloom also supports flash-attn (optional). To enable it, please install flash-attn (v2.6.3 is used during training) manually and set os.environ['DISABLE_FLASH_ATTN'] = "0" in infer.py:8
```
- model-name: Specify model version, see the model cards (eg: songbloom_full_150s/songbloom_full_150s_dpo);
- local-dir: Dir where the weights and config files are downloaded;
- input-jsonl: input raw data;
- output-dir: Dir where the output audio saved;
- n-samples: How many audios will be generated for each input term;
## Mac Silicon
Set these environment variables before running:
```
export PYTORCH_ENABLE_MPS_FALLBACK=1
export DISABLE_FLASH_ATTN=1
```
When loading the model, explicitly pass the MPS device and use float32, not bfloat16:
```
import torch
device = torch.device('mps')
model = SongBloom_Sampler.build_from_trainer(cfg, strict=False, dtype=torch.float32, device=device)
```
## Citation
```
@article{yang2025songbloom,
title={SongBloom: Coherent Song Generation via Interleaved Autoregressive Sketching and Diffusion Refinement},
author={Yang, Chenyu and Wang, Shuai and Chen, Hangting and Tan, Wei and Yu, Jianwei and Li, Haizhou},
journal={arXiv preprint arXiv:2506.07634},
year={2025}
}
```
## License
SongBloom (codes and weights) is released under the [LICENSE](LICENSE).