Home

Awesome

AI Powered Notes app [Sample]

This sample is a simple note taking app that uses local APIs and models to provide AI powered features. The app is built using WinUI3.

image

Watch the Build session: Use AI for "Real Things" in your Windows apps

Set Up

You will need to have Visual Studio installed with the latest workloads for WinAppSDK and WinUI 3 development. You can find instructions on how to set up your environment here.

Clone the repository and open the solution in Visual Studio. Before you can get started exploring the sample, you will need to download the ML model files required for the project and place them in the onnx-models folder.

The final folder structure should look like this:

Folder Structure

Downloading Phi3

The model can be downloaded from the following link:

Huggingface models are in repositories which you can clone to get the model files. Clone the Phi3 model repository and copy the required files to this project.

Phi-3-mini-4k-instruct-onnx has 3 different versions inside it's repo. We are using the DirectML versions in this project. Copy the contents of the directml/directml-int4-awq-block-128 folder to a new folder called phi-3-directml-int4-awq-block-128 under onnx-models folder.

Downloading all-MiniLM-L6-v2

The model can be downloaded from the following link:

This is model we use for semantic search. The two files you will need are model.onnx and vocab.txt. Create a new folder under onnx-models called embedding and place the files there.

Downloading Sliero VAD

The Sliero Voice Activity Detection model can be downloaded from the following link:

This is the model we use for smart chunking of audio and the only file you will need is the sliero_vad.onnx file.

This should also be placed under a new folder called whisper under the onnx-models folder.

Downloading Whisper

The process for getting the Whisper model is a bit more involved, as it needs to be manually generated with Olive.

This can all be done from the command line and only requires Python as a dependency, to get your model, follow these steps:

  1. Clone the Olive repository and navigate to the Whisper example folder:
git clone https://github.com/microsoft/Olive
cd Olive/examples/whisper
  1. Install Olive from source and use a virtual environment or conda:
pip install git+https://github.com/microsoft/Olive
  1. Install the required packages:
python -m pip install -r requirements.txt
pip install onnxruntime
pip install onnxruntime_extensions
  1. Prepare the Whisper model
python prepare_whisper_configs.py --model_name openai/whisper-small --multilingual --enable_timestamps
  1. Run the Olive workflow to generate the optimized model
olive run --config whisper_cpu_int8.json --setup
olive run --config whisper_cpu_int8.json

  1. The generated model will be in the \models\conversion-transformers_optimization-onnx_dynamic_quantization-insert_beam_search-prepost folder.

  2. Rename the model from whisper_cpu_int8_cpu-cpu_model.onnx to whisper_small.onnx and place it in the onnx-models/whisper folder.

Troubleshooting

TextRecognition APIs are not available

The TextRecognition APIs are not yet available in the public release of the WindowsAppSDK and their usage is commented out in this sample. We will update the app to use those APIs once they are available publicly.

Path name too long

You might run into an issue if you clone the repo in a location that will make the path too long to some of the generated binaries. Recomendation is to place the repo closer to the root of the drive and rename the repo folder name to something shorter. Alternatively, you can change the settings in Windows to support long paths https://learn.microsoft.com/en-us/windows/win32/fileio/maximum-file-path-limitation?tabs=registry#enable-long-paths-in-windows-10-version-1607-and-later .

Olive version and config mismatch

You might run into an issue where the config file and olive version is not compatible. Because Olive is in rapid development, you will need to ensure the version of the config file matches the olive version you want to use. To ensure the latest versions of the examples can be run without issues, you have to install Olive from source. Alternatively, for downstream dependencies, we suggest pinning the Olive version that worked for you.

Contributing

This project welcomes contributions and suggestions. Most contributions require you to agree to a Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us the rights to use your contribution. For details, visit https://cla.opensource.microsoft.com.

When you submit a pull request, a CLA bot will automatically determine whether you need to provide a CLA and decorate the PR appropriately (e.g., status check, comment). Simply follow the instructions provided by the bot. You will only need to do this once across all repos using our CLA.

This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact opencode@microsoft.com with any additional questions or comments.

Trademarks

This project may contain trademarks or logos for projects, products, or services. Authorized use of Microsoft trademarks or logos is subject to and must follow Microsoft's Trademark & Brand Guidelines. Use of Microsoft trademarks or logos in modified versions of this project must not cause confusion or imply Microsoft sponsorship. Any use of third-party trademarks or logos are subject to those third-party's policies.