A game device issues a notification that the player is prompted to input a voice. After the notification is issued, voice data representing a voice signal, having a predetermined time length, which is inputted to a voice input element, is repeatedly acquired. Furthermore, each time the voice data is acquired, it is determined whether or not the acquired voice data satisfies a predetermined selecting condition. Thereafter, only voice data, which is determined to satisfy the selecting condition, is stored as selected voice data. The game device outputs, when a game image showing a game character speaking is displayed on the display, a sound effect representing a voice of the game character by using at least a portion of the selected voice data.