README.md

---
license: cc-by-4.0
language:
- en
- it
- pt
- de
- fr
- es
- ja
- zh
tags:
- automatic-speech-recognition
- speech
- audio
- Transformer
- flow-matching
- discrete-flow-matching
- pytorch
- hf-asr-leaderboard
---

# Drax: Speech Recognition with Discrete Flow Matching

## Model Overview

The Drax model family provides speech recognition models based on discrete flow matching. 
The `drax-v1` model supports eight languages: English, Spanish, French, Portuguese, German, Italian, Japanese and Chinese.
It is an encoder-decoder model consists of a Whisper-large-v3 encoder, and a DiT based decoder, with a total of ~1.2B parameters.

More details on usage in our GitHub repo, [https://github.com/aiola-lab/drax](https://github.com/aiola-lab/drax) and our [paper](https://arxiv.org/abs/2510.04162).

## Usage

See [https://github.com/aiola-lab/drax](https://github.com/aiola-lab/drax) for installation instructions.

```python
from drax import Transcriber

asr = Transcriber(model_path="aiola/drax-v1")
result = asr.transcribe("/path/to/audio.wav", language="en")
print(result[0].transcript)
```

Control sampling steps, temperature etc.

```python
from drax import Transcriber

asr = Transcriber(model_path="aiola/drax-v1")
result = asr.transcribe("/path/to/audio.wav", language="en", sampling_steps=32, temperature=1e-2)
print(result[0].transcript)
```

Batch inference:

```python
from drax import Transcriber

asr = Transcriber(model_path="aiola/drax-v1")
audio_paths = ["/path/to/audio1.wav", "/path/to/audio2.wav"]
languages = ["en", "de"]
result = asr.transcribe(audio_paths, language=languages)
print(result.transcript)
```

## Citation

```bibtex
@article{navon2025drax,
  title={Drax: Speech Recognition with Discrete Flow Matching},
  author={Navon, Aviv and Shamsian, Aviv and Glazer, Neta and Segal-Feldman, Yael and Hetz, Gill and Keshet, Joseph and Fetaya, Ethan},
  journal={arXiv preprint arXiv:2510.04162},
  year={2025}
}
```