python/FunASR-XL.git

			@@ -10,55 +10,68 @@


			### Steps:
			1. Download the whole directory (`funasr/runtime/python/onnxruntime`) to the local.
			2. Install the related packages.
			```bash
			pip install requirements.txt
			```
			3. Download the model.
			- [Download Link](https://swap.oss-cn-hangzhou.aliyuncs.com/zhifu.gzf/export/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/model.onnx?OSSAccessKeyId=LTAI4FxMqzhBUx5XD4mKs296&Expires=2036094510&Signature=agmtMkxLEviGg3Rt3gOO4PvfrJY%3D)
			- Put the model into the `resources/models`.
			```text
			.
			├── demo.py
			├── rapid_paraformer
			│ ├── __init__.py
			│ ├── kaldifeat
			│ ├── __pycache__
			│ ├── rapid_paraformer.py
			│ └── utils.py
			├── README.md
			├── requirements.txt
			├── resources
			│ ├── config.yaml
			│ └── models
			│ ├── am.mvn
			│ ├── model.onnx # Put it here.
			│ └── token_list.pkl
			├── test_onnx.py
			├── tests
			│ ├── __pycache__
			│ └── test_infer.py
			└── test_wavs
			├── 0478_00017.wav
			└── asr_example_zh.wav
			```
			4. Run the demo.
			1. Export the model.
			- Command: (`Tips`: torch 1.11.0 is required.)

			```shell
			python -m funasr.export.export_model [model_name] [export_dir] [true]
			```
			`model_name`: the model is to export.

			`export_dir`: the dir where the onnx is export.

			More details ref to ([export docs](https://github.com/alibaba-damo-academy/FunASR/tree/main/funasr/export))

			- `e.g.`, Export model from modelscope
			```shell
			python -m funasr.export.export_model 'damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch' "./export" true
			```
			- `e.g.`, Export model from local path, the model'name must be `model.pb`.
			```shell
			python -m funasr.export.export_model '/mnt/workspace/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch' "./export" true
			```


			2. Install the `rapid_paraformer`.
			- Build the rapid_paraformer `whl`
			```shell
			git clone https://github.com/alibaba/FunASR.git && cd FunASR
			cd funasr/runtime/python/onnxruntime
			python setup.py bdist_wheel
			```
			- Install the build `whl`
			```bash
			pip install dist/rapid_paraformer-0.0.1-py3-none-any.whl
			```

			3. Run the demo.
			- Model_dir: the model path, which contains `model.onnx`, `config.yaml`, `am.mvn`.
			- Input: wav formt file, support formats: `str, np.ndarray, List[str]`
			- Output: `List[str]`: recognition result.
			- Example:
			```python
			from rapid_paraformer import RapidParaformer
			from rapid_paraformer import Paraformer

			model_dir = "/nfs/zhifu.gzf/export/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch"
			model = Paraformer(model_dir, batch_size=1)

			config_path = 'resources/config.yaml'
			paraformer = RapidParaformer(config_path)
			wav_path = ['/nfs/zhifu.gzf/export/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/example/asr_example.wav']

			wav_path = ['test_wavs/0478_00017.wav']

			result = paraformer(wav_path)
			result = model(wav_path)
			print(result)
			```

			## Speed

			Environment：Intel(R) Xeon(R) Platinum 8163 CPU @ 2.50GHz

			Test [wav, 5.53s, 100 times avg.](https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/asr_example_zh.wav)

			\| Backend \| RTF \|
			\|:-------:\|:-----------------:\|
			\| Pytorch \| 0.110 \|
			\| Onnx \| 0.038 \|


			## Acknowledge
			1. We acknowledge [SWHL](https://github.com/RapidAI/FunASR) for contributing the onnxruntime(pthon api).
			1. We acknowledge [SWHL](https://github.com/RapidAI/RapidASR) for contributing the onnxruntime(python api).