Each of a plurality of TG (tone generation) units includes a plurality of TG elements corresponding to a plurality of tone pitches. A plurality of resonant boxes are provided in corresponding relation to the TG units, and each box is formed of a material having a characteristic different from the materials forming the other boxes. One or more of the TG elements to be driven for tone generation are determined on the basis of tone pitch designating information and tone color setting information, and the determined TG elements are driven to vibrate. Because the resonant boxes are formed of different materials, tones can be generated with peculiar tone colors in accordance with the materials of the resonant boxes corresponding to the vibrated TG elements. In an alternative, the TG elements in each of the TG units may be formed of a material of a characteristic different from materials forming the TG elements of the other TG units.