Abstract: The determination of the number of the simultaneously observed objects is an important problem of the MUSIC algorithm. As the result of the covariance matrix eigendecomposition the eingevector and unitary matrix are obtained. The unitary matrix defines the transformation of the measurement vector. Consequently the obtained vector is orthogonal to the noise subspace. The solution of the eigendecomposition problem can be avoided assuming a-priori knowledge of the unitary matrix. Practically, the passive sonar detects the first appeared noisy object. This enable to beforehand define the set of the unitary matrixes exactly matched to the previously determined directions of the signal arrival. The article presents the form of the matrix and features of the measurement vector transformation. The results of the DOA algorithm computer simulation for passive sonar composed of four hydrophones developed in the OBR CTM S.A. are also presented.