top of page

Research Works

I am involved in writing code for quality assessment of AI generated music, researching AI generated music metrics.

  • Designed and improved several harmonic and rhythmic metrics based on music theory to assess the quality of AI generated music, with respect to sequence structure, harmony and sound effect of a MIDI piece.

​

​

  • Music MIDI copyright detection algorithms

    • Implemented through Levenshtein metrics using C++, intergrated with Python using PyBind.

    • Several improvements was made to boost the performance from original 10 hours to 8 minutes.

  • Recently studying Deep-learning based music generation algorithms 

    • Mainly onTransformer-derived models including Transformer XL and GC Transformer.

bottom of page