Secure narrowband digital conferencing
J. Bertrand · 2002
The use of single speaker models, such as linear predictive coding (LPC), for narrowband secure communication considerably increases the difficulty of secure conferencing which is achieved via a conference director that enforces the single speaker model. Recently developed speech processing algorithms can process speech resulting from multiple speakers and retain sufficient intelligibility and quality. How these algorithms can be used to support a full-duplex secure conferencing system is described. The proposed method uses one of the recently developed analysis-by-synthesis schemes, such as multipulse or code-excited linear-prediction (CELP), as the voice processing algorithm. The conferencing scheme is based on a conference director that is connected to all participants via a regular FDX phone link.>