Estimation of the Number of Sources Present in Instantaneous and Anechoic Mixtures

In this paper we propose a method for estimating the number of sources in the instantaneous and anechoic mixing systems without the knowledge of the system parameters. From the observation of the two or more mixing signals, we can obtain the scattering plot that consists of the phase and amplitude information of the mixing signals. Using the sparsity property of the sources in the time-frequency domain and the time-frequency coefficients of the sources form individual clusters that are resolvable, the number of sources present in a given mixed signals can be estimated by estimating the number of clusters present in the scattering plot. A criterion used for the estimation of the number of clusters is also presented in this paper. The results are shown for both synthetic and experimental (real) mixing data. From the results it can be seen that the estimation of the number of sources is 100% accurate for the examples provided.