---
language: zh
license: apache-2.0
tags:
  - automatic-speech-recognition
  - ASR
  - chinese
  - speech
---

## 📂 Project Tree

```
WSChuan-ASR
├── paraformer_large_chuan/
│ ├── config.yaml
│ ├── model.pt
│ └── infer.py
│
├── Qwen2.5-omni3B/
| ├──added_tokens.json
| ├──args.json
| ├──char_template.jinja
| ├──config.json
| ├──generation_config.json
| ├──merges.txt
| ├──model-00001-of-00003.safetensors
| ├──model-00002-of-00003.safetensors
| ├──model-00003-of-00003.safetensors
| ├──model.safetensors.index.json
| ├──preprocessor_config.json
| ├──special_tokens_map.json
| ├──spk_dict.pt
| ├──tokenizer_config.json
| ├──tokenizer.json
| ├──video_preprocessor_config.json
| └──vocab.json
│
├── .gitattributes
└── README.md
```

## ASR Leaderboard
| Model | Model Size | WSC-Eval-ASR - Easy | WSC-Eval-ASR - Hard | WSC-Eval-ASR - Total | Magicdata - Conversation | Magicdata - Daily-Use | Avg. |
| --- | --- | --- | --- | --- | --- | --- | --- |
| **with LLM** |  |  |  |  |  |  |  |
| Kimi-Audio<sup></sup> | 7B | 16.65 | 28.66 | 17.66 | 24.67 | **5.77** | 18.68 |
| FireRedASR-LLM<sup></sup> | 8.3B | 12.80 | 25.27 | 14.40 | 17.68 | 6.69 | 15.37 |
| Qwen2.5-omni<sup></sup> | 3B | 16.94 | 26.01 | 18.20 | 20.40 | 6.32 | 17.69 |
| Qwen2.5-omni-WSC-Finetune⭐ | 3B | 14.36 | 24.14 | 15.61 | 18.45 | 6.15 | 15.74 |
| <span style="background-color: #d4edda; padding: 0 2px;">Qwen2.5-omni+internal data⭐</span> | 3B | 13.17 | 23.36 | 14.81 | 18.50 | 5.88 | 15.14 |
| <span style="background-color: #d4edda; padding: 0 2px;">Qwen2.5-omni-WSC-Finetune + internal data⭐</span> | 3B | 12.93 | 23.19 | 14.25 | 17.95 | <u>5.89</u> | 14.84 |
| **without LLM** |  |  |  |  |  |  |  |
| SenseVoice-small<sup></sup> | 234M | 17.43 | 28.38 | 18.39 | 23.50 | 8.77 | 19.29 |
| Whisper<sup></sup> | 244M | 52.06 | 63.99 | 53.59 | 55.88 | 52.03 | 55.51 |
| FireRedASR-AED<sup></sup> | 1.1B | 13.29 | 23.64 | 14.62 | 17.84 | 6.69 | 15.14 |
| Paraformer<sup></sup> | 220M | 14.34 | 24.61 | 15.66 | 19.81 | 8.16 | 16.52 |
| Paraformer-WSC-Finetune⭐ | 220M | 12.15 | 22.60 | 13.51 | 16.60 | 8.02 | 14.58 |
| <span style="background-color: #d4edda; padding: 0 2px;">Paraformer + internal data⭐</span> | 220M | <u>11.93</u> | <u>21.82</u> | <u>13.14</u> | <u>15.61</u> | 6.77 | <u>13.85</u> |
| <span style="background-color: #d4edda; padding: 0 2px;">Paraformer-WSC-Finetune + internal data</span>⭐ | 220M | **11.59** | **21.59** | **12.87** | **14.59** | 6.28 | **13.38** |



## ASR Inference
### Paraformer_large_Chuan
```
export CUDA_VISIBLE_DEVICES=7
root_dir=./test_data
test_sets=("WSC-Eval-ASR" "WSC-Eval-ASR-Hard" "WSC-Eval-ASR-Easy")
model_dir=./model_dir

out_rootdir=./results
mkdir -p $out_rootdir
python infer_paraformer.py \
  --model $model_dir \
  --wav_scp_file $root_dir/$test_data/wav.scp \
  --output_dir $out_rootdir/debug \
  --device "cuda" \
  --output_file $out_dir/hyp.txt
```
---

### Qwen2.5-Omni-3B_Chuan
```
python infer_qwen2.5omni.py \
    --wavs_path /path/to/your/wav.scp \
    --out_path /path/to/your/results.txt \
    --gpu 0 \
    --model /path/to/your/model
```


## Note:

Original: https://huggingface.co/ASLP-lab/WSChuan-ASR

Duplicated by: https://huggingface.co/spaces/huggingface-projects/repo_duplicator