Welcome to my IT masters thesis project. The goal of this project was to detect stress in human voice using techniques like:
- sound processing
- machine learning
The general idea is as follows:
First you generate a trained model with the algorithm used in options (Options -> Options).
Next you need to a find model which performs the best on training data.
Then you can save the trained model as a weka model file on a disk (Action -> Generate Weka model).
Later you can use trained model to classify some new wav file (File -> Classify file).
You can provide a pretrained weka model and a sample wav file to classify.
You can also save weka ARFF file to perform experiments using different algorithms in Weka GUI.
- Java
- Maven
mvn clean
mvn package
To start the application please run the following command:
java -jar bin/vsa-1.0.jar
- Please use mono WAV files 44100 Hz only