Speech recognition technology is developing due to the improvement of recording technology and the advent of deep learning technology. In the game, various works using voice recognition have been released, starting with the family computer that appeared in the 1980s.In this research, we have developed a system that changes the reaction and response of NPCs by inferring emotions from the acoustic information spoken by the player. For emotion analysis, we used “UserLocal”, which is a voice recognition sentiment analysis web API, and implemented it using the game engine Unity.