Measuring the emotional qualities of an album using Spotify Track Audio Features with lyrical sentiment analysis via RoBERTa. The data is gathered,
processed, and merged in a .py file and then used in a Jupyter Notebook to build a summary presentation.
You can install any of the required libraries using pip like in the example below.
pip install pandas
You will also need access to the Spotify and Genius APIs.
Any suitable model can be used from 🤗 Hugging Face with some modifications but j-hartmann/emotion-english-distilroberta-base is used in this case.
The album used is Power, Corruption & Lies by New Order. The original track list is used with these eight tracks:
- Age of Consent
- We All Stand
- The Village
- 5 8 6
- Your Silent Face
- Ultraviolence
- Ecstasy
- Leave Me Alone