Towards the Implementation of an Automatic Composition System for Popular Songs

Lyrics are the most recognizable part of a popular song, and corresponding singing voice makes it easy to convey the emotion and experience inside the lyrics to the listeners. However, most current automatic composition systems that can generate pop music lack the consideration for lyrics and singing voice. In this paper, we propose an Integrated architecture of Automatic composition system concerned with Singing voice from given lyrics for popular songs (IAmSinging). With the help of a three-stage pipeline, score, performance and audio files can be generated for users easily. To evaluate the feasibility and availability, we implemented a user-friendly web-based application named CCOM AI-composition system based on this architecture. The experimental results indicates that the system can generate satisfactory results and could be an enjoyable solution for musical amateurs.