Keshav Bhandari PhD StudentEmail: k.bhandari@qmul.ac.uk Website: https://keshavbhandari.github.io/ ProfileResearchProfileProject title: Neuro-Symbolic Automated Music Composition Abstract: Recent developments in symbolic music generation have explored various token encoding schemes such as byte pair encoding and relative positional encoding in conjunction with transformer architectures. This has led to impressive results in composing complex multi-track and multi-instrument music with coherence, incorporating diverse range of styles, polyphony and harmonic progressions. However, symbolic generators lack critical structure and direction necessary for long-term coherence. Furthermore, their ability to stimulate emotions and adhere to the rules of musical counterpoint is still limited and actively being researched. Through a PhD at the UKRI Centre for Doctoral Training in AI and Music at Queen Mary University, I plan to address these core limitations and push our boundaries to generate music with AI to the next level. C4DM theme affiliation: Music Informatics / Music GenerationResearchResearch Interests:Music Generation, Audio Synthesis, Music Information Research