Abstract: A method of recording visual information such as words of a song in an audio recording medium and reproducing the same is provided. Picture information obtained through a TV camera is quantized into pulse signals. Using these pulse signals, a sine wave signal having an audio frequency band is modulated. Further, synchronizing signals are added to the modulated signal at predetermined intervals to form a character signal. Said character signal is recorded in one of the two channels of an audio recording medium where as a musical accompaniment is recorded in another channel. When played, such character signal is reproduced on a display screen such that a user can read the words to sing the song to the simultaneously reproduced accompaniment.