AI/wekws
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
wekws/runtime/android/README.md
veelion 184e8a8da4
[runtime/android] Add how to run on Android, and change onnxruntime to 1.12.1 (#111) 
* add Usage

* LookupCustomMetadataMap() is deprecated in 1.13.1, so change to 1.12.1

Co-authored-by: weiliang <weiliang.chong@day-care.cn>
2022-11-24 17:17:43 +08:00

824 B
Raw Blame History

Usage

Most of AI engineers are not familiar with Android development, this is a simple ‘how to’.

  1. Train your model with your data

  2. Export pytorch model to onnx model

  3. Convert onnx model for mobile deployment

python -m onnxruntime.tools.convert_onnx_models_to_ort your-model.onnx

you will get your-model.ort and your-model.with_runtime_opt.ort

  1. Install Android Studio and open path of wekws/runtime/android and build

NOTE: The default feature_dim in code is 40, if your model’s is 80, change it here ./app/src/main/cpp/wekws.cc

  feature_config = std::make_shared<wenet::FeaturePipelineConfig>(40, 16000);  // 40 -> 80

It’s also can be built on Linux by runing bash ./gradlew build

  1. Install app/build/outputs/apk/debug/app-debug.apk to your phone and try it.