Provided is a speaker box, which includes a first damping member fixed to the housing and covering the sound outlet, and a second damping member fixed in the receiving space and fixed to the housing. The second damping member, the base and/or the cover define an extension cavity in communication with the front acoustic cavity. When the speaker produces sound, the extension cavity stabilizes frequency response variation caused by variation in the dynamic height of the air space. When the diaphragm of the speaker is at the lowest and highest positions of the air space, the variation in the final acoustic frequency response of the speaker box is always small, and the final acoustic frequency response of the speaker is also small. Thus, the intermodulation distortion of speakers in mobile devices or other thin devices can be reduced, and a sound quality of the speaker box can be improved.