Audio Conferencing Using Dynamic Channel-Assignment System

摘要

This study examines dynamic channel assignment techniques to demonstrate their feasibility in audio teleconferencing. Computer simulation of dynamic assignment is used to investigate the rules for controlling the bit-rate and is used to subjectively evaluate the speech quality of audio conferencing system. Efficient encoding of speech signals requires the removal of redundancies in the speech waveform. Predictive codings techniques are applicable to dynamic channel assignment. Adaptive Predictive Coding at 16 kbps is suitable for a high quality / high rate channel. The closely related technique of Linear Predictive Coding is appropriate for a low quality / low rate channel. A dual-mode coder combines both the above techniques together with a switching control algorithm. The control algorithm selects the mode of the dual-mode coder based on the relative amplitudes of the conference participants. Computer simulations of a dual-mode coder indicate that the technique is well suited to audio conferencing. Dynamic assignment of the coding mode introduces only minor quality degradations when proper control algorithms are used. The overall impression is one of high quality coding for most of the duration of the conference.