mugaggamichael/Luganda-KWS

This is our Fourth year project. # Luganda-KWS This is our Fourth year project. The aim of this project is to create a keyword spotting system that can run on resource constricted systems for a group of keywords. The intention of this is to be able to identify a group of keywords in the Luganda language and also carry out these commands on any system this application runs on. A dataset we worked with has been uploaded to this link. drive.google.com Below is the procedure to undertake when setting up the application 1.Place all the cleaned dataset in a folder, Dataset. Make sure each keyword is under its own folder for example Dataset\Kuba 2.Run the prepare_dataset.py file 3.A data.json file should populate containing the MFCCs, labels and for the keywords present in the dataset