To reproduce voice information with high sound quality without drilling any hole through a casing by fixing a magnet to the casing having a vibration area capable of vibration via a frame to generate a magnetic field to a voice coil that is fixed to a 2nd end of a pole piece whose 1st end is fixed to the vibration area.

One of both ends of a pole piece 35 is fixed by a bolt 36 almost at the center of a vibration area 30 which is formed thinner than other areas of a 2nd casing 28, and a bobbin where a voice coil 40 is wound is fixed at the other end of the piece 35. The coil 40 interlinks to a magnetic field formed between the 1st and 2nd members 46 and 50 made of ferromagnetic materials by a magnetic circuit 44 consisting of the members 46 and 50 and a magnet 48 that is fixed onto the inner surface of the member 46. Then the member 46 of the circuit 44 is fixed to one of both ends of a frame 52 with the other end of the frame 52 fixed to an area other than the area 30 by means of a bolt 54 and a nut 56.