In this paper, we propose a WORD-oriented technique for encrypting (and decrypting) digital audio files based on rotation and XOR operations. The key-concepts of the designed encryption algorithm are the RX (Rotation-XOR) operations, i.e., the plain audio samples are first left-rotated by the sum-of-digits of the previous audio samples, and then XOR-ed with the previous audio samples. The designed encryption algorithm encodes a digital audio file into a random (noise) like audio file, from human visual as well as statistical points of view. Several encryption and decryption evaluation metrics, such as spectrogram, adjacent sample correlation coefficient, number of sample change rate, peak signal-to-noise ratio, etc., are applied on several digital audio files of varying sizes in order to empirically assess the performance and efficiency of the proposed technique. The results of these metrics validate the robustness of the designed technique.