---
license: apache-2.0
language:
- en
pipeline_tag: text-to-speech
---

<h1 align="center">DiFlow-TTS: Compact and Low-Latency Zero-Shot Text-to-Speech with Discrete Flow Matching</h1>

<p align="center">
  <a href="https://github.com/Fsoft-AIC/DiFlowTTS/tree/main"><img src="https://img.shields.io/badge/GitHub-Code-181717?logo=github" alt="GitHub"></a>
  <a href="https://arxiv.org/abs/2509.09631"><img src="https://img.shields.io/badge/arXiv-2509.09631-b31b1b?logo=arxiv" alt="Paper"></a>
  <a href="https://fsoft-aic.github.io/SonNN45-Demo/projects/diflowtts/"><img src="https://img.shields.io/badge/Demo-Page-blue?logo=googlechrome&logoColor=white" alt="Demo"></a>
  <a href="https://interspeech2026.org/en-AU"><img src="https://img.shields.io/badge/Interspeech-2026-orange" alt="Interspeech 2026"></a>
</p>

> [!NOTE]
> DiFlow-TTS is trained on 470 hours of the LibriTTS dataset, which consists of **predominantly neutral speech**. As a result, **it may not perform well on prompts with strong emotional expression**.

Download [DiFlow-TTS](https://huggingface.co/Fsoft-AIC/DiFlowTTS/blob/main/diflow-tts.ckpt) checkpoint, and place it as follows:

```
root/
└── ckpts/
    └── diflow-tts.ckpt
```