In this paper, we address the security and efficiency issues of two recently proposed audio scrambling schemes. We show that these two audio scrambling schemes are actually vulnerable against various attacks such as ciphertext-only attack, known-plaintext attack and chosen-plaintext attack. We also demonstrate that one of these two schemes is lack of efficiency in terms of generating the key stream using the dynamic password generator (DPG). Furthermore, we briefly discuss the ways to improve the security and efficiency of these two audio scrambling schemes.