Eurovision Song Contest Dataset

The Eurovision Song Contest is a freely-available dataset containing metadata, contest ranking and voting data of 1562 songs that have competed in the Eurovision Song Contests. The upcoming release will also contain audio features.

Every year, the dataset is updated with the contest's results. This release contains the contestant metadata, contest ranking and voting data of 1562 entries that participated in the Eurovision Song Contest from its first occurrence in 1956 until now. The corresponding audio for every song can be streamed through YouTube.

The metadata and voting data are provided by the EurovisionWorld fansite.

Downloading the dataset

The dataset can be downloaded here. To replicate it, follow the instructions at the bottom of the readme.

Audio

With the contestants.csv in the same folder as the audio.py file, the YouTube audio streams of all songs can be collected by running python3 audio.py. Alternatively, sh run.sh audio or sh run.sh docker audio can be used to scrape locally or use a Docker container to scrape the streams.

Using the dataset in your research paper?

Please contact janne [dot] spijkervet [at] gmail [dot] com

How to get started

To get an initial idea of the dataset, an example Jupyter Notebook is created in the examples directory. This can be opened with jupyter notebook. To replacite the dataset, see below or:

Easy setup

The run.sh file makes it easy to either replicate the full dataset or download the latest version and extract the audio features from all the songs. By default, sh run.sh will run the scraper from the local Python environment.

Run sh run.sh docker to build the Dockerfile and run the main.py from within the container. No additional setup should be necessary. This will replicate the dataset, both the contestants.csv and votes.csv files.

The audio can be scraped from either within or outside the Docker container:

sh run.sh docker audio
sh run.sh audio

Audio Features

The audio features can be extracted once all the audio is present in the audio folder using:

sh audio_features.sh

This will launch a Docker container with Essentia's stream music extractor installed. Alternatively, audio_features.py can be run given Essentia's extractor is installed in the PATH environment.

Data description

The competition ranking is provided for both finals and semi-finals. The country-to-country voting data contains 47007 voting activities, and is separated by jury- and televoting after it was introduced in 2016.

contestants.csv

votes.csv

Replication

It is recommended to use Docker by running sh run.sh docker, or use a local installation by just invoking sh run.sh. To also obtain the audio, run either sh run.sh audio or sh run.sh docker audio.

To replicate the dataset, a WebDriver for either Chrome, Firefox, or Safari is required, e.g. the WebDriver for Chrome, along with the Selenium Python package (pip3 install selenium). Follow the instructions to setup the WebDriver here. The project's dependencies can be installed using:

pip3 install -r requirements.txt

Use the following command to extact the data of all Eurovision Song Contests between 1956 and 2019:

python3 main.py --start 1956 --end 2019

This will create a contestants.csv and votes.csv file.