Discoveries of interacting relational patterns between listeners and recordings within Open Dataset of Audio Quality

This repository contains the code and data for the paper "Discoveries of interacting relational patterns between listeners and recordings within Open Dataset of Audio Quality".

Dataset Information

The dataset used in this study is the Open Dataset of Audio Quality (ODAQ), which contains audio recordings and listener ratings. The dataset is publicly available and can be accessed here. The additional data from Ball State University (BSU) can also be found here. Details about the dataset can be found in the original paper's repository here.

The datasets are already included in this repository under the Data\ODAQ and ODAQ_v1_BSU directories.

Project Setup Instructions

After cloning the repository, you should create a virtual environment and install the required packages. You can do this by running the following commands:

1. Move to the project folder

cd ODAQ_CEDA  # or the name of your cloned folder

2. Create a virtual environment and activate it

For Windows:

python -m venv venv
venv\Scripts\activate

For MacOS and Linux:

python3 -m venv venv
source venv/bin/activate

3. Install required packages

pip install -r requirements.txt

Script Usage

1. ODAQ_competition_ranking_experts_students.py and ODAQ_kmeans_ranking_experts_students.py

   These scripts are used to analyze the ODAQ dataset and perform ranking based on expert and student ratings. They can be run directly after setting up the environment and installing the required packages. The scripts will generate output files containing the results of the analysis as well as visualizations such as heatmaps, clustermaps, and spaghetti plots. The intermediate results (ranking and clustering) are saved in the Results directory as .pkl files.

2. contingency_table.py

   This script is used to create a contingency table from the ODAQ dataset. It processes the data and generates a table that summarizes the interactions between listeners and recordings. While the table in the paper displays frequencies, this script allows you to see which recordings and listeners belong to each grouping. The contingency tables are output as CSV files in the Results directory.