The Machine Intelligence and Music Informatics Research Group is a specialised research team within the Department of Computer Science.
Music Informatics includes the study of computational models of music analysis, music generation, and music performance. Interests of the Music Informatics group include machine learning, computational musicology, audio signal processing, music knowledge representation, and applications. The group is also interested in wider aspects of machine learning on signals and symbolic (discrete) data, such as bioacoustics and industrial signal and knowledge analysis, and the novel application of techniques from these areas to music.
- Deep learning for music signal analysis:
- automatic music transcription,
- multi-pitch detection,
- onset detection,
- instrument identification.
- Challenges in deep neural network learning
- sequential structure recognition and generation
- abstract pattern recognition
- design of inductive biases
- Machine learning for music analysis, classification, music generation and style emulation
- Algorithms for music segmentation, voice segregation, hierarchical structuring, and analysis
- Computational musicology
- Music knowledge representation
- Scientific and industrial applications of signal processing and machine learning.