## Using funasr with libtorch
[FunASR](https://github.com/alibaba-damo-academy/FunASR) hopes to build a bridge between academic research and industrial applications on speech recognition. By supporting the training & finetuning of the industrial-grade speech recognition model released on ModelScope, researchers and developers can conduct research and production of speech recognition models more conveniently, and promote the development of speech recognition ecology. ASR for Fun!
### Steps:
1. Export the model.
- Command: (`Tips`: torch >= 1.11.0 is required.)
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 --model-name damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch --export-dir ./export --type torch --quantize False
```
- `e.g.`, Export model from local path, the model'name must be `model.pb`.
```shell
python -m funasr.export.export_model --model-name ./damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch --export-dir ./export --type torch --quantize False
```
2. Install the `funasr_torch`.
install from pip
```shell
pip install --upgrade funasr_torch -i https://pypi.Python.org/simple
```
or install from source code
```shell
git clone https://github.com/alibaba/FunASR.git && cd FunASR
cd funasr/runtime/python/libtorch
pip install -e ./
```
3. Run the demo.
- Model_dir: the model path, which contains `model.torchscripts`, `config.yaml`, `am.mvn`.
- Input: wav formt file, support formats: `str, np.ndarray, List[str]`
- Output: `List[str]`: recognition result.
- Example:
```python
from funasr_torch 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)
wav_path = ['/nfs/zhifu.gzf/export/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/example/asr_example.wav']
result = model(wav_path)
print(result)
```
## Performance benchmark
Please ref to [benchmark](https://github.com/alibaba-damo-academy/FunASR/blob/main/funasr/runtime/python/benchmark_libtorch.md)
## 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 (FP32) |
|:--------:|:----------:|
| Pytorch | 0.110 |
| Libtorch | 0.048 |
| Onnx | 0.038 |
## Acknowledge
This project is maintained by [FunASR community](https://github.com/alibaba-damo-academy/FunASR).
Raw data
{
"_id": null,
"home_page": "https://github.com/alibaba-damo-academy/FunASR.git",
"name": "funasr-torch",
"maintainer": "",
"docs_url": null,
"requires_python": "",
"maintainer_email": "",
"keywords": "funasr,paraformer, funasr_torch",
"author": "Speech Lab, Alibaba Group, China",
"author_email": "funasr@list.alibaba-inc.com",
"download_url": "https://files.pythonhosted.org/packages/a3/c8/6c21fe6dfa610990d05a71ee759d19960e6be4675c12403f9e5e955caafb/funasr_torch-0.0.4.tar.gz",
"platform": "Any",
"description": "## Using funasr with libtorch\n\n[FunASR](https://github.com/alibaba-damo-academy/FunASR) hopes to build a bridge between academic research and industrial applications on speech recognition. By supporting the training & finetuning of the industrial-grade speech recognition model released on ModelScope, researchers and developers can conduct research and production of speech recognition models more conveniently, and promote the development of speech recognition ecology. ASR for Fun\uff01\n\n\n### Steps:\n1. Export the model.\n - Command: (`Tips`: torch >= 1.11.0 is required.)\n\n More details ref to ([export docs](https://github.com/alibaba-damo-academy/FunASR/tree/main/funasr/export))\n\n - `e.g.`, Export model from modelscope\n ```shell\n python -m funasr.export.export_model --model-name damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch --export-dir ./export --type torch --quantize False\n ```\n - `e.g.`, Export model from local path, the model'name must be `model.pb`.\n ```shell\n python -m funasr.export.export_model --model-name ./damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch --export-dir ./export --type torch --quantize False\n ```\n\n\n2. Install the `funasr_torch`.\n \n install from pip\n ```shell\n pip install --upgrade funasr_torch -i https://pypi.Python.org/simple\n ```\n or install from source code\n\n ```shell\n git clone https://github.com/alibaba/FunASR.git && cd FunASR\n cd funasr/runtime/python/libtorch\n pip install -e ./\n ```\n\n3. Run the demo.\n - Model_dir: the model path, which contains `model.torchscripts`, `config.yaml`, `am.mvn`.\n - Input: wav formt file, support formats: `str, np.ndarray, List[str]`\n - Output: `List[str]`: recognition result.\n - Example:\n ```python\n from funasr_torch import Paraformer\n\n model_dir = \"/nfs/zhifu.gzf/export/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch\"\n model = Paraformer(model_dir, batch_size=1)\n\n wav_path = ['/nfs/zhifu.gzf/export/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/example/asr_example.wav']\n\n result = model(wav_path)\n print(result)\n ```\n\n## Performance benchmark\n\nPlease ref to [benchmark](https://github.com/alibaba-damo-academy/FunASR/blob/main/funasr/runtime/python/benchmark_libtorch.md)\n\n## Speed\n\nEnvironment\uff1aIntel(R) Xeon(R) Platinum 8163 CPU @ 2.50GHz\n\nTest [wav, 5.53s, 100 times avg.](https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/asr_example_zh.wav)\n\n| Backend | RTF (FP32) |\n|:--------:|:----------:|\n| Pytorch | 0.110 |\n| Libtorch | 0.048 |\n| Onnx | 0.038 |\n\n## Acknowledge\nThis project is maintained by [FunASR community](https://github.com/alibaba-damo-academy/FunASR).\n",
"bugtrack_url": null,
"license": "The MIT License",
"summary": "FunASR: A Fundamental End-to-End Speech Recognition Toolkit",
"version": "0.0.4",
"split_keywords": [
"funasr",
"paraformer",
" funasr_torch"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "f83cfe1c89e7a7ab68848d8eb7f1f6d45ecb55d8a939cecbc3f400747c93f365",
"md5": "01dc738634e2f67f9737f9a826f615f5",
"sha256": "8276af8bfba9d6344c03d7c67750080123a496268e0889b06712ba4dab08dff7"
},
"downloads": -1,
"filename": "funasr_torch-0.0.4-py3-none-any.whl",
"has_sig": false,
"md5_digest": "01dc738634e2f67f9737f9a826f615f5",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": null,
"size": 2465,
"upload_time": "2023-04-14T15:21:23",
"upload_time_iso_8601": "2023-04-14T15:21:23.004740Z",
"url": "https://files.pythonhosted.org/packages/f8/3c/fe1c89e7a7ab68848d8eb7f1f6d45ecb55d8a939cecbc3f400747c93f365/funasr_torch-0.0.4-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "a3c86c21fe6dfa610990d05a71ee759d19960e6be4675c12403f9e5e955caafb",
"md5": "d3762156f3a20dc2fe03d569a33f846d",
"sha256": "41785e73674326ff147551f09b8b64efe3d0b6d6ddc23cef3d46388b3582423e"
},
"downloads": -1,
"filename": "funasr_torch-0.0.4.tar.gz",
"has_sig": false,
"md5_digest": "d3762156f3a20dc2fe03d569a33f846d",
"packagetype": "sdist",
"python_version": "source",
"requires_python": null,
"size": 2544,
"upload_time": "2023-04-14T15:21:25",
"upload_time_iso_8601": "2023-04-14T15:21:25.109963Z",
"url": "https://files.pythonhosted.org/packages/a3/c8/6c21fe6dfa610990d05a71ee759d19960e6be4675c12403f9e5e955caafb/funasr_torch-0.0.4.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2023-04-14 15:21:25",
"github": true,
"gitlab": false,
"bitbucket": false,
"github_user": "alibaba-damo-academy",
"github_project": "FunASR.git",
"lcname": "funasr-torch"
}