Creating Singing Vocal Expressions by means of Interactive Evolutionary Computation

Today, researches for singing by computer have attracted attention. VOCALOID is an application to realize that aim. By inputing lyrics and melody, users can make songs sung by the computer. In order to make the singing voice more "human", users must control frequency curve very carefully. Comparing with inputing lyrics or melody, this controlling presents heavy overhead for users. In this research, we propose a system for easily optimizing frequency curves. This system searches for parameters with a type of GA called Interactive Evolutionary Computation (IEC). On the other hand, the system using IEC has a phase for users to evaluate, we need to consider the tiredness of users. This tiredness is connected to the effectiveness of the search with GA. In this research, for the analysis of the tiredness of users, we evaluated the convergence performance of GA to fit the goal which is known in advance. As a result, we found that our method has better convergence performance than a previous method.