A flat speaker unit is provided herein. The flat speaker unit includes a first porous electrode, a second porous electrode, and a vibrating membrane with an electret layer disposed there between. In one embodiment, a plurality of supporting members may be configured between the vibrating membrane and the first porous electrode, or between the vibrating membrane and the second porous electrode. In one embodiment, a flat speaker device is provided with at least two flat speaker unit stacked together. By electrically connecting two ends of a signal source respectively to the first and second porous electrodes, or, in another embodiment, electrically connecting one end of the signal source to both of the first and second porous electrodes and connecting another end of the signal source to the vibrating membrane, a sound with low THD is generated accordingly from the flat speaker unit.