---
title: Mega-ASR-bf16
canonical_url: "https://www.modelscope.cn/models/mlx-community/Mega-ASR-bf16"
md_url: "https://www.modelscope.cn/models/mlx-community/Mega-ASR-bf16.md"
repository: mlx-community/Mega-ASR-bf16
last_updated: 2026-05-23
license: apache-2.0
model_type:
  - mega_asr
base_model:
  - zhifeixie/Mega-ASR
  - Qwen/Qwen3-ASR-1.7B
base_model_relation: finetune
parameters: 2.0B
tensor_type:
  - BF16
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - pytorch
language:
  - en
  - zh
downloads: 41
stars: 0
tags:
  - mlx
  - speech-to-text
  - asr
  - robust-asr
  - qwen3-asr
---

# Mega-ASR-bf16

> Mega-ASR-bf16 - mlx-community 在 ModelScope 开源的模型。This model was converted to MLX format from zhifeixie/Mega-ASR (built on Qwen/Qwen3-ASR-1.7B) using mlx-audio.

mlx-community/Mega-ASR-bf16 是 ModelScope 魔搭社区上的 2.0B 参数机器学习模型，采用 apache-2.0 许可，基于 zhifeixie/Mega-ASR、Qwen/Qwen3-ASR-1.7B 构建。

- **Repository**: mlx-community/Mega-ASR-bf16
- **License**: apache-2.0
- **Parameters**: 2.0B
- **Base model**: zhifeixie/Mega-ASR, Qwen/Qwen3-ASR-1.7B
- **Tags**: mlx, speech-to-text, asr, robust-asr, qwen3-asr
- **Downloads**: 41
- **Stars**: 0
- **Last updated**: 2026-05-23

Source: https://www.modelscope.cn/models/mlx-community/Mega-ASR-bf16

---

# Mega-ASR-bf16

This model was converted to MLX format from [`zhifeixie/Mega-ASR`](https://huggingface.co/zhifeixie/Mega-ASR) (built on [`Qwen/Qwen3-ASR-1.7B`](https://huggingface.co/Qwen/Qwen3-ASR-1.7B)) using [mlx-audio](https://github.com/Blaizzy/mlx-audio).

Mega-ASR is a **robustness layer over Qwen3-ASR-1.7B**: a tiny audio-quality **router** classifies each utterance as clean or degraded and switches a dense **LoRA adapter** in/out of the base weights at inference — degraded audio runs the LoRA (robust) path, clean audio runs the unmodified base path. This recovers large WER gains on noisy/far-field speech while leaving clean-speech accuracy unchanged.

> The base weights are stored as **dense bf16** on purpose: Mega-ASR adds fp32 LoRA deltas to the base at inference, so the base cannot be quantized without losing the runtime router/LoRA switching.

## Use with mlx-audio

```bash
pip install mlx-audio
```

```python
from mlx_audio.stt import load

model = load("mlx-community/Mega-ASR-bf16")
result = model.generate("audio.wav", language="en")
print(result.text)
```

CLI:

```bash
python -m mlx_audio.stt.generate --model mlx-community/Mega-ASR-bf16 --audio audio.wav
```

The router decides per-utterance automatically; no flags needed.

## Validation

Reproduces the paper's published robustness gains. Word Error Rate on the real **NOIZEUS** corpus (8 noise types × 4 SNR × 30 utterances, Apple Silicon):

| SNR | base (Qwen3-ASR) | Mega-ASR (robust) | paper base | paper robust |
|---|---:|---:|---:|---:|
| 0 dB | 23.35 | 20.61 | 23.97 | 19.80 |
| 5 dB | 8.47 | 6.51 | — | — |
| 10 dB | 3.31 | 2.17 | 3.41 | 2.79 |
| 15 dB | 2.12 | 0.83 | — | — |
| **overall** | **9.31** | **7.53** | **9.45** | **7.52** |

Overall robust WER **7.53 vs the paper's 7.52** — a ~20% relative reduction over the Qwen3-ASR baseline, reproduced. On clean read speech (FLEURS) the model matches plain Qwen3-ASR, as intended.

## License & attribution

Apache-2.0. Built on [zhifeixie/Mega-ASR](https://huggingface.co/zhifeixie/Mega-ASR) (adapter + router) and [Qwen/Qwen3-ASR-1.7B](https://huggingface.co/Qwen/Qwen3-ASR-1.7B) (base).