Introduction

Automatic music generation is a task that generates music melodies / accompaniment with computer. Usually, a dataset containing multiples pieces of music is used for training a model of artificial neuron networks. The resulting model is then used to generate music with an input of random noise / dedicated meta -data.

Currently, there are many challenging issues in generated music such as the lack of music structure (due to the difficulties on modelling long -term dependency in music), the existence of inharmonic in music (due to the difficulties on modelling relationship between notes in music) and the uncertainty in music style (as the model fails to capture temporal relationship between music events representing music style). You are now required to write a one-page essay (excluding reference list) that discusses the possible development of automatic music generation  methods  in  near  future. The  proposed  development  in  the  essay must be supported by a review of existing literatures. Please note, it is not necessary to address technical details in your essay.

If you are have no ideas on the specific developments to address in your essay, you may choose one of the following candidate topics:

⚫     Generating music with music structure

⚫     Generating waveform based singing voice with expressiveness

⚫     Using a single model to generate music in different styles

⚫     Detecting computer generated music (either symbolic music or waveform based music) However, you are encourage to propose your own topic for this coursework.