Disclosed is a speaker configured to reduce or eliminate distortions when a diaphragm thereof vertically moves in sound reproduction and which can be downsized. This speaker includes: a cylindrical frame having a bottom face and an open top; a magnetic circuit disposed on a inner bottom face of the frame; a voice coil having a coil portion thereof disposed in a magnetic gap of the magnetic circuit; the diaphragm disposed in an opening of the frame, and fixed to the voice coil along an inner periphery thereof, and to the frame via a first edge along an outer periphery thereof; and a support fixed to the diaphragm on the side of the magnetic circuit at one end and to the vicinity of the magnetic circuit via the second edge on the other end. The first edge is substantially analogous to the second edge in shape with respect to the boundary therebetween.