Creation of Ideal User's Voice Using User's own UTAU Voice and Interactive Genetic Algorithm
Asami Inoue, Kota Nomura, Makoto Fukumoto · 2018
Since many books and papers on voice quality improvement are circulating around the world, we think that there are at least those who want to change their own voice. However, there is an image that voice quality improvement is difficult. The reason is that it is difficult to objectively evaluate user's oneself with voice training by user's oneself, and user cannot improve voice quality user's oneself easily. Therefore, in this research, we propose and verify the method of creating user's ideal voice, which is the goal when improving voice quality based on user's voice by using software UTAU and Interactive Genetic Algorithm (IGA). As results of experiment, in a search experiment using IGA, the average and the maximum evaluation values increased significantly. Furthermore, using the best individuals found in this search by comparing significant increase was also observed (p<;0.05). Therefore, it turns out that the proposed method has the possibility of making the ideal voice of the user.