README.md 2.01 KB
Newer Older
chenxj's avatar
chenxj committed
1
2
3
# asr_onnx
## 论文

chenxj's avatar
chenxj committed
4
## 模型结构
chenxj's avatar
chenxj committed
5
![image](https://developer.hpccube.com/codes/modelzoo/asr_onnx/-/raw/main/resources/silero_stt_model.jpg)
dcuai's avatar
dcuai committed
6
## 算法原理
chenxj's avatar
chenxj committed
7
8
9
10
![image](https://developer.hpccube.com/codes/modelzoo/asr_onnx/-/raw/main/resources/asr.png)
## 数据集

## 环境配置
chenxj's avatar
chenxj committed
11
12
[光源](https://www.sourcefind.cn/#/service-details)可拉取推理的docker镜像,在[光合开发者社区](https://cancon.hpccube.com:65024/4/main/)可下载onnxruntime安装包。asr_onnx推荐的镜像如下:
```
shantf's avatar
shantf committed
13
14
15
16
17
18
19
20
docker pull image.sourcefind.cn:5000/dcu/admin/base/pytorch:2.1.0-ubuntu20.04-dtk24.04.1-py3.10
cd asr_onnxruntime #进入当前项目目录
docker run -d -t --privileged --device=/dev/kfd --device=/dev/dri/ --network=host --group-add video -v /opt/hyhal:/opt/hyhal:ro -v `pwd`:/mnt --name=asr-test image.sourcefind.cn:5000/dcu/admin/base/pytorch:2.1.0-ubuntu20.04-dtk24.04.1-py3.10
docker exec -it asr-test /bin/bash
cd /mnt
pip install onnx -i https://pypi.tuna.tsinghua.edu.cn/simple
pip install pysoundfile -i https://pypi.tuna.tsinghua.edu.cn/simple

chenxj's avatar
chenxj committed
21
```
chenzk's avatar
chenzk committed
22
23
24
25
下载模型 (https://models.silero.ai/models/en/en_v5.onnx) 到当前目录,建立wavs文件夹添加测试wav文件。

预训练权重快速下载中心:[SCNet AIModels](http://113.200.138.88:18080/aimodels) ,项目中的预训练权重可从快速下载通道下载:[en_v5](http://113.200.138.88:18080/aimodels/findsource-dependency/weight/-/raw/main/en_v5.onnx)

chenxj's avatar
chenxj committed
26
## 推理
chenxj's avatar
chenxj committed
27
28
```
python3 main.py --model_dir="./en_v5.onnx" --wav_dir="./wavs/" --warmup=1
shantf's avatar
shantf committed
29
# --wav_dir:需要推理的语音路劲,如"./speech_orig.wav";speech_orig.wav是文件夹中已经存在的语音
chenxj's avatar
chenxj committed
30
```
dcuai's avatar
dcuai committed
31
## result
chenxj's avatar
chenxj committed
32
33
34
35
36
37
38
39
![image](https://developer.hpccube.com/codes/modelzoo/asr_onnx/-/raw/main/resources/asr_result.png)
### 精度
暂无
## 应用场景
### 算法类别
语音识别
### 热点应用行业
交通,金融,医疗,教育,家居
chenxj's avatar
chenxj committed
40
41
## 源码仓库及问题反馈
https://developer.hpccube.com/codes/modelzoo/asr_onnx
dcuai's avatar
dcuai committed
42
## 参考资料
chenxj's avatar
chenxj committed
43
* [silero-models](https://github.com/snakers4/silero-models)
sugon_cxj's avatar
sugon_cxj committed
44