Teaching

Music Data Mining

I have taught Music Data Mining several times at Indiana University. The course is an introductory deep learning and machine learning course for music, audio, and text data analysis. Students learn how to analyze audio and text data, build music information retrieval (MIR) models, and evaluate MIR systems. The course uses tools such as Librosa, Scikit-learn, and Keras.

In my teaching, I focus on explaining complex ideas in clear and accessible ways. I start with small, concrete examples so that students develop an intuitive and solid understanding before applying these ideas to larger systems and real-world problems. Examples of my lecture materials is available here: Week 02 Audio Content Analysis Librosa and Week 07 Deep_Learning_Basic.

The course includes lab sessions with hands-on programming, followed by assignments that reinforce the concepts introduced in the lectures and labs. For the final project, students design and implement an MIR system and present their work to the class.

Course Topics (example from Fall 2023)

  • Music and Audio Content Analysis using Librosa
  • Linear Regression | Music Mood Regression
  • Logistic Regression | Music Mood Classification
  • Classification and Evaluation | MIREX
  • Support Vector Machine
  • Deep Learning Basics | FMA
  • Convolutional Neural Networks | Large-scale Music Genre Classification
  • Recurrent Neural Networks
  • Annotation and Voice Analysis

I am currently revising and adapting this course for the UIUC iSchool, aligning it with my recent research directions. This course is closely connected to my research on MIR. Selected relevant publications include.

  • Kahyun Choi, Automatic Music Composition Algorithm using Regression and an Emotion Plane, 2007
  • Kahyun Choi, Performance Improvement of Music Mood Classification Using Hyper Music Features, 2010
  • Kahyun Choi, Computational Lyricology: Quantitative Approaches to Understanding Song Lyrics and Their Interpretations, 2018
  • Kahyun Choi and Minje Kim, “A Comparative Analysis of Poetry Reading Audio: Singing, Narrating, or Somewhere In Between?” IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), 2024
  • Kahyun Choi, “Word Embedding-Based Text Complexity Analysis,” in Proceedings of the 19th iConference, 2024
  • Kahyun Choi, “Bimodal Music Subject Classification via Context-Dependent Language Model,”  in Proceedings of the 16th iConference, 2021.
  • Kahyun Choi and J. Stephen Downie, “A Trend Analysis on Concreteness of Popular Song Lyrics,”  in Proceedings of the 6th International Conference on Digital Libraries for Musicology (DLfM), 2019.
  • Kahyun Choi, Jin Ha Lee, Xiao Hu, and J. Stephen Downie, “Music Subject Classification Based on Lyrics and User Interpretations,” Proceedings of the American Society for Information Science and Technology (ASIS&T), 2016.
  • Kahyun Choi, Jin Ha Lee, Craig Willis, J. Stephen Downie, “Topic Modeling Users’ Interpretations of Songs to Inform Subject Access in Music Digital Libraries,” In Proceedings of the IEEE/ACM Joint Conference in Digital Libraries (JCDL), 2015.
  • Kahyun Choi, Jin Ha Lee, and J. Stephen Downie, “What is this song about anyway?: Automatic classification of subject using user interpretations and lyrics,” In Proceedings of the IEEE/ACM Joint Conference on Digital Libraries (JCDL), 2014.
  • Haven Kim and Kahyun Choi, “LyCon: Lyrics Reconstruction from the Bag-of-Words Using Large Language Models,” the 18th International Society for Music Information Retrieval Conference (ISMIR), 2024
  • Xiao Hu, Kahyun Choi, J. Stephen Downie, “A Framework for Evaluating Multimodal Music Mood Classification,” Journal of the Association for Information Science and Technology (JASIST), vol. 68, no. 2, pp. 273-285, 2017.
  • Xiao Hu, Kahyun Choi, Yun Hao, Sally Jo Cunningham, Jin Ha Lee, Audrey Laplante, David Bainbridge, and J. Stephen Downie, “Exploring the Music Library Association Mailing List: A Text Mining Approach,” in Proceedings of the 18th International Society for Music Information Retrieval Conference (ISMIR), 2017
  • Xiao Hu, Jin Ha Lee, David Bainbridge, Kahyun Choi, Peter Organisciak, and J. Stephen Downie, “The MIREX Grand Challenge: a framework of holistic user experience evaluation in music information retrieval,” Journal of the Association for Information Science and Technology (JASIST), vol. 68, no. 1, pp. 97-112, 2015.
  • Jin Ha Lee, Xiao Hu, Kahyun Choi, and J. Stephen Downie, “MIREX Grand Challenge 2014 on User Experience: Qualitative analysis of user feedback”, in Proceedings of the 16th International Society for Music Information Retrieval Conference (ISMIR), 2015.
  • J. Stephen Downie, Xiao Hu, Jin Ha Lee, Kahyun Choi, Yun Hao, and Sally Jo. Cunningham, “Ten Years of MIREX (Music Information Retrieval Evaluation eXchange): Reflections, Challenges and Opportunities,” in Proceedings of the International Society of Music Information Retrieval Conference (ISMIR), Taipei, Taiwan, Oct. 2014. 
  • Xiao Hu, Jin Ha Lee, Kahyun Choi, and J. Stephen Downie, “A Cross-Cultural Study of Mood in K-POP Songs,” in Proceedings of the International Society of Music Information Retrieval Conference (ISMIR), Taipei, Taiwan, Oct. 2014. 
  • Jin Ha Lee, Kahyun Choi, Xiao Hu, and J. Stephen Downie, “K-Pop Genres: A Cross-Cultural Exploration”, in Proceedings of the International Society of Music Information Retrieval Conference (ISMIR), Curitiba, Brazil, Oct. 2013.