---
title: HotBias
canonical_url: "https://www.modelscope.cn/models/coming245/HotBias"
md_url: "https://www.modelscope.cn/models/coming245/HotBias.md"
repository: coming245/HotBias
last_updated: 2026-11-02
license: "Apache License 2.0"
pipeline_tag: auto-speech-recognition
tasks:
  - auto-speech-recognition
library_name:
  - pytorch
frameworks:
  - pytorch
domain:
  - audio
downloads: 0
stars: 0
tags:
  - hotword-biasing
  - contextual-biasing
  - cross-backbone-transfer
---

# HotBias

> HotBias - coming245 在 ModelScope 开源的模型。Model weights for HotBias: Hotword Biasing for ASR with Cross-Domain Generalization and Cross-Model Transfer.

coming245/HotBias 是 ModelScope 魔搭社区上的auto-speech-recognition模型，采用 Apache License 2.0 许可。

- **Repository**: coming245/HotBias
- **License**: Apache License 2.0
- **Tasks**: auto-speech-recognition
- **Tags**: hotword-biasing, contextual-biasing, cross-backbone-transfer
- **Downloads**: 0
- **Stars**: 0
- **Last updated**: 2026-11-02

Source: https://www.modelscope.cn/models/coming245/HotBias

---

# HotBias

Model weights for **HotBias: Hotword Biasing for ASR with Cross-Domain Generalization and Cross-Model Transfer**.

**Official code and usage:** [QwenAudio/FunResearch/HotBias](https://github.com/QwenAudio/FunResearch/tree/main/HotBias).

HotBias adds gated, adaptive hotword biases to frozen ASR models. A pretrained biasing core can be transferred to another recognizer by aligning target-specific input projections.

## Checkpoints

| File | Purpose |
| --- | --- |
| `source_transfer.pt` | Pretrained HotBias core and source input projections, exported from the Qwen3-based source system. Use it to initialize target alignment. |
| `nano_aligned.pt` | Aligned HotBias weights for Fun-ASR-Nano-2512. |
| `firered_aligned.pt` | Aligned HotBias weights for FireRedASR2-AED. |

These files contain HotBias parameters. Obtain the corresponding ASR backbone and tokenizer separately. All three checkpoints share the same frozen biasing core.

## Training data

The HotBias module was trained on speech data from **AISHELL-2**, and generated hotword candidate lists.

## Download and use

```python
from modelscope import snapshot_download

model_dir = snapshot_download('coming245/HotBias')
print(model_dir)
```

Pass `nano_aligned.pt` or `firered_aligned.pt` from the downloaded directory to the corresponding inference script. For example, from the HotBias code directory:

```bash
python fun_asr_nano/infer.py \
  --model-dir /path/to/Fun-ASR-Nano-2512 \
  --nano-model-code /path/to/Fun-ASR/model.py \
  --checkpoint /path/to/hotbias-checkpoints/nano_aligned.pt \
  --audio /path/to/audio.wav \
  --hotword '池州' --hotword '赤州'
```

See the [GitHub README](https://github.com/QwenAudio/FunResearch/tree/main/HotBias) for installation, FireRed inference, and target alignment. See [checkpoint details](https://github.com/QwenAudio/FunResearch/blob/main/HotBias/docs/checkpoints.md) for compatibility and [reproduction notes](https://github.com/QwenAudio/FunResearch/blob/main/HotBias/docs/reproduction.md) for evaluation scope.

## File integrity

From the downloaded directory, run:

```bash
sha256sum -c SHA256SUMS
```