A music sound generation apparatus generates music-sound data based on sound source data, and outputs the music-sound data in synchronization with audio data obtained separately from the music-sound data. A channel assigner assigns a predetermined number of channels to the audio data and the music-sound data. A read control unit reads the audio data and the music-sound data in accordance with their assigned channels. A signal processing unit generates predetermined music by controlling the sound level of the read audio data and the read music-sound data on each assigned channel and mixing the audio data and the music-sound data. When detecting no unoccupied channel, the channel assigner determines, as an unoccupied channel, a channel selected from channels excluding the channel to which the audio data has been assigned. The signal processing unit performs sound elimination on the unoccupied channel.