Steganography is a technique of concealing secret information within an innocuous-looking carrier signal. Speech steganography has become increasingly important in a variety of applications, including military, satellite, and mobile communications. However, conventional methods have failed to meet the maximum-security requirements, resulting in reduced robustness and imperceptibility performance. In this paper, a novel speech steganography mechanism is proposed to address these issues by using shift invariant continuous wavelet transform (SI-CWT) for data hiding. The message data is hidden in the low-level speech bands using pseudo noise sequences, resulting in a steganography output speech signal. To recover the original speech and message signal, a zero-crossing rate (ZCR)-based speech activity and message detection (SAMD) mechanism is applied on the receiver side. The proposed steganography system is evaluated using various performance metrics, including robustness, imperceptibility, and security. Simulation results demonstrate that the proposed system outperforms conventional methods in terms of these metrics, indicating that it is a promising solution for secure speech communication. Additionally, potential limitations and weaknesses of the proposed system are discussed, highlighting the need for further research in this area. Overall, this novel speech steganography mechanism has significant potential for enhancing the security and reliability of speech communication in various applications.