A singing voice synthesizing method and a singing voice synthesizing apparatus in which the singing voice is synthesized using performance data such as MIDI data. The performance data entered is analyzed as the musical information of the sound pitch, sound duration and the lyric (S2, S3). From the analyzed music information, the lyric is accorded to a string of sounds to form singing voice data (S5, S6). The speech waveform of the singing voice is formulated from the singing voice data (S7, S8). The waveform of the music sound is formulated from the input performance data (S14). The portion of the performance data used for the singing voice is desirably not used in reproducing the music sound, or lowered in the reproducing sound volume. A program, a recording medium and a robot apparatus, in which the singing voice is synthesized from performance data, are also disclosed.