An ultrasonic speaker including: a vibration film having a conductive layer; a pair of fixed electrodes provided oppositely to respective surfaces of the vibration film; and an ultrasonic transducer that causes the vibration film to generate sound waves by applying a DC bias voltage to the conductive layer of the vibration film while applying an alternating signal between the pair of fixed electrodes, and outputs the sound waves generated by the vibration film from two sound wave output surfaces via through-holes provided in each of the pair of fixed electrodes. The two sound wave output surfaces of the ultrasonic transducer are disposed in a direction intersecting with a sound wave emitting direction at right angles.