About

Short BIO

Hi I'm Katerina Kosta, a senior machine learning researcher at Speech, Audio and Music Intelligence (SAMI) team at ByteDance/TikTok. I pursued my PhD from the Centre for Digital Music, Queen Mary University of London, conducting research on modelling dynamic variations in expressive music performance. Other research interests include custom data structures, pattern recognition and machine learning for music synthesis and analysis of perceived emotion in music audio.

I received degrees from National and Kapodistrian University of Athens (Mathematics) and Filippos Nakas Conservatory, Athens (Piano), and a Sound and Music Computing Masters from the Music Technology Group, UPF, Barcelona.

Publications

2022

  • Kosta, K., Lu, W. C., Medeot, G., Chanquion, P. (2022). A deep learning method for melody extraction from a polyphonic symbolic music representation. In Proceedings of the 23rd International Society for Music Information Retrieval Conference (ISMIR), pp. 757-763.
    • PDF
  • Zhang, D., Wang, J-C., Kosta, K., Smith, J. B. L., Zhou, S. (2022). Modeling the rhythm from lyrics for melody generation of pop songs. In Proceedings of the 23rd International Society for Music Information Retrieval Conference (ISMIR).
    • PDF

2021

  • Micchi, G., Kosta, K., Medeot, G., Chanquion, P. (2021). A deep learning method for enforcing coherence in automatic chord recognition. In Proceedings of the 22nd International Society for Music Information Retrieval Conference (ISMIR), pp. 443-451.
    • PDF
  • Medeot, G., Cherla, S., Kosta, K., McVicar, M., Abdallah, S., Selvi, M., Newton-Rex, E., & Webster, K. (2021). A method of generating music data. US Patent App. 16/967,064.
    • PDF

2018

  • Medeot, G., Cherla, S., Kosta, K., McVicar, M., Abdallah, S., Selvi, M., Newton-Rex, E., & Webster, K. (2018). StructureNet: Inducing Structure in Generated Melodies. In Proceedings of the 19th International Society for Music Information Retrieval Conference (ISMIR), pp. 725-731, Paris, France.
    • PDF
  • Kosta, K., O. F. Bandtlow, E. Chew (2018). Dynamics and relativity: Practical implications of dynamic markings in the score. Journal of New Music Research, 47(5), pp. 438-461.
    • PDF
  • Kosta, K., O. F. Bandtlow, E. Chew (2018). MazurkaBL: Score-aligned loudness, beat, and expressive markings data for 2000 Chopin Mazurka recordings. In Proceedings of the 4th International Conference on Technologies for Music Notation and Representation (TENOR), pp. 85-94, Montreal, Canada.
    • PDF

2017

  • Kosta, K., O. F. Bandtlow, E. Chew (2017). Dynamic change points in music audio capture dynamic markings in score. 18th International Society for Music Information Retrieval Conference (ISMIR), Late-Breaking and Demo Session, Suzhou, China.
    • PDF
  • Kosta, K. (2017). Computational Modelling and Quantitative Analysis of Dynamics in Performed Music. Ph.D. Thesis. Centre for Digital Music, Queen Mary University of London, London, UK.
    • PDF

2016

  • Kosta, K., R. Ramirez, O. F. Bandtlow, E. Chew (2016). Mapping between dynamic markings and performed loudness: A machine learning approach. Journal of Mathematics and Music, 10(2): 149-172.
    • PDF
  • Kosta, K., O. F. Bandtlow, E. Chew (2016). Outliers in Performed Loudness Transitions: An Analysis of Chopin Mazurka Recordings. In Proceedings of the 14th International Conference for Music Perception and Cognition (ICMPC), pp. 601-604, July 5-9, 2016, San Francisco, California, USA.
    • PDF

2015

  • Kosta K., R. Ramirez, O. F. Bandtlow, E. Chew (2015). Predicting loudness levels and classifying dynamic markings in recorded music. In Proceedings of 8th International Workshop on Machine Learning and Music (MML2015), Machine Learning for Music Generation, Vancouver, Canada.
    • PDF
  • Kosta, K., O. F. Bandtlow, E. Chew (2015). A Change-point Approach Towards Representing Musical Dynamics. In T. Collins, D. Meredith, A. Volk (eds.): Mathematics and Computation in Music: 5th International Conference, MCM 2015, London, UK, June 22-25, 2015, Proceedings, pp. 179-184, Lecture Notes in Computer Science 9110, Berlin: Springer.
    • PDF

2014

  • Kosta, K., Li, S. (2014). 2013 Performance Studies Network International Conference. Computer Music Journal, 38(2): 78-80.
    • PDF
  • Kosta, K., O. F. Bandtlow and E. Chew (2014). A Study of Score Context-dependent Dynamics in Piano Performance. In Proceedings of the Performance Studies Network International Conference (PSN3), Jul 17-20, Cambridge, UK.
    • PDF
  • Kosta, K., O. F. Bandtlow, E. Chew (2014). Practical Implications of Dynamic Markings in the Score: Is piano always piano? In Proceedings of the 53rd Audio Engineering Society (AES) Meeting on Semantic Audio, Jan 26-29, London, UK.
    • PDF

2013

  • Kosta, K., Y. Song, G. Fazekas, M. Sandler (2013). A Study of Cultural Dependence of Perceived Mood in Greek Music. In Proceedings of the 14th International Society for Music Information Retrieval (ISMIR), pp. 317-322, Nov 4-8, Curitiba, Brazil.
    • PDF

2012

  • Kosta, K., M. Marchini, H. Purwins (2012). Unsupervised Chord-Sequence Generation from an Audio Example. In Proceedings of the 13th International Society for Music Information Retrieval (ISMIR), pp. 481-486, Porto, Portugal.
    • PDF

2011

  • Kosta, K. (2011). Unsupervised Generation of Chord Sequences from a Sound Example. Masters thesis, Music Technology Group, Universitat Pompeu Fabra, Barcelona, Spain.
    • PDF
Get in Touch

Contact

katkost@gmail.com