EmotionGUI-UoA

Summer Research Project at the University of Auckland

Demo Video

Introduction

EmotionGUI is a user-intuitive tool for the visualization and annotation of emotions in speech and video signals. The tool is based on the valence-arousal two-dimensional emotion model from psychology and enables multiple users to annotate emotions in a simple and efficient way. EmotionGUI now consists of three sections: visualize, annotate, and Live-Audio. Moreover, the tool supports multiple machine learning models that can be loaded into the tool to predict valence-arousal values and visualize them. This research aims to address the need for a user-intuitive tool that can facilitate annotation by multiple users and is open-sourced for future development. The GUI is built on top of PyQt5.

Visualize

The visualize section allows users to see the emotions associated with a speech signal on a two-dimensional plane by inputting corresponding valence and arousal values.

Annotation

The annotation section allowsusers to mark the valence and arousal values of the speech or video signal as a function of time.

Live-Audio

The new Live-Audio section provides a real-time audio recording feature that can help create more quality synthetic data for different emotional training and annotation tasks.

Installing from source

• Write git clone https://github.com/kaustubh-s1/EmotionGUI-UoA.git in your terminal in any directory.
• Open the folder in VS code or any other text-editor of your choice.
• Create a virtual environment by typing-

  • On mac:

    python3 -m virtualenv venv
    source venv/bin/activate
    

  • On Windows:

    virtualenv venv
    venv\Scripts\activate
    

  • On Linux:

    virtualenv venv
    virtualenv -p /usr/bin/python3 venv
    

After creating the virtual environment install all the dependencies by typing:- pip install -r requirements.txt in the editor's terminal. You're good to go. Open the main.py file and run it. Or in the editor's terminal type python main.py The GUI screen will be opened and you can start now.

If in case any dependency does not get installed properly try installing it using conda for which you might want to install Anaconda

Using the GUI with an exe (Only supported for Windows as of now)

• Download this zip file
• Now go to the directory EmotionGUI-UoA/output/main/main.exe
• Open main.exe and get started