Hugging Face's logo

declare-lab
/
TangoFlux

Text-to-Audio
Inference Endpoints
Model card Files Community
TangoFlux
File size: 2,886 Bytes 
3d7b891
 
 
 
714699f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
52e8e5d
714699f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
 
---
license: mit
datasets:
- cvssp/WavCaps
---

<h1 align="center">✨ 
<br/>  
TangoFlux: Super Fast and Faithful Text to Audio Generation with Flow Matching and Clap-Ranked Preference Optimization 
<br/>
✨✨✨


</h1>

<div align="center">
  <img src="https://github.com/declare-lab/TangoFlux/blob/main/assets/tf_teaser.png" alt="TangoFlux" width="1000" />

<br/>

<div style="display: flex; gap: 10px; align-items: center;">
  <a href="https://openreview.net/attachment?id=tpJPlFTyxd&name=pdf">
    <img src="https://img.shields.io/badge/Read_the_Paper-blue?link=https%3A%2F%2Fopenreview.net%2Fattachment%3Fid%3DtpJPlFTyxd%26name%3Dpdf" alt="arXiv">
  </a>
  <a href="https://huggingface.co/declare-lab/TangoFlux">
    <img src="https://img.shields.io/badge/TangoFlux-Huggingface-violet?logo=huggingface&link=https%3A%2F%2Fhuggingface.co%2Fdeclare-lab%2FTangoFlux" alt="Static Badge">
  </a>
  <a href="https://tangoflux.github.io/">
    <img src="https://img.shields.io/badge/Demos-declare--lab-brightred?style=flat" alt="Static Badge">
  </a>
  <a href="https://huggingface.co/spaces/declare-lab/TangoFlux">
    <img src="https://img.shields.io/badge/TangoFlux-Huggingface_Space-8A2BE2?logo=huggingface&link=https%3A%2F%2Fhuggingface.co%2Fspaces%2Fdeclare-lab%2FTangoFlux" alt="Static Badge">
  </a>
  <a href="https://huggingface.co/datasets/declare-lab/CRPO">
    <img src="https://img.shields.io/badge/TangoFlux_Dataset-Huggingface-red?logo=huggingface&link=https%3A%2F%2Fhuggingface.co%2Fdatasets%2Fdeclare-lab%2FTangoFlux" alt="Static Badge">
  </a>
  <a href="https://github.com/declare-lab/TangoFlux">
    <img src="https://img.shields.io/badge/Github-brown?logo=github&link=https%3A%2F%2Fgithub.com%2Fdeclare-lab%2FTangoFlux" alt="Static Badge">
  </a>
</div>



</div>

## Model Overview

TangoFlux consists of FluxTransformer blocks which are Diffusion Transformer (DiT) and Multimodal Diffusion Transformer (MMDiT), conditioned on textual prompt and duration embedding to generate audio at 44.1kHz up to 30 seconds. TangoFlux learns a rectified flow trajectory from audio latent representation encoded by a variational autoencoder (VAE). The TangoFlux training pipeline consists of three stages: pre-training, fine-tuning, and preference optimization. TangoFlux is aligned via CRPO which iteratively generates new synthetic data and constructs preference pairs to perform preference optimization.


## Getting Started

## Citation

```bibtex
@article{Hung2025TangoFlux,
  title = {TangoFlux: Super Fast and Faithful Text to Audio Generation with Flow Matching and Clap-Ranked Preference Optimization},
  author = {Chia-Yu Hung and Navonil Majumder and Zhifeng Kong and Ambuj Mehrish and Rafael Valle and Bryan Catanzaro and Soujanya Poria},
  year = {2025},
  url = {https://openreview.net/attachment?id=tpJPlFTyxd&name=pdf},
  note = {Available at OpenReview}
}
```