In this paper, the experimental investigation of coherently combined air-modulated speakers is achieved. This paper addresses a basic architecture for coherent-combined air-modulated speakers by means of active phase control, and then the fundamental theory for coherent combination using stochastic parallel gradient descent algorithm is described. Numerical simulation on coherently combining sound sources with horn is carried out. Moreover, an experiment of far-field coherently combining two air-modulated speakers has been conducted, and the configuration scheme of parameters in practical terms is presented. Experimental results show that when the algorithm converges, the sound pressure level at the test spot is 4 dB higher than that of single source. A power spectrum analysis indicates that the sound power of coherently combining case is close to the sum of the sound powers between the coherently combining of the fundamental frequency components and the non-coherently combining of the other frequency components for all sound sources. Consequently, it is revealed that the acoustic wave phases for air-modulated speakers can be effectively controlled by the active phase control algorithm, and an evident coherent-combining effect is obtained.