Adaptive gain control for speech quality improvement and echo suppression

Speech quality in hands-free telephone sets can be improved and acoustic echos can be suppressed by gain control in the transmission path of such telephone systems. The gain factor is computed as a function of the short-term level of the input signal, which allows the separation of speech from background noise and can cope with different speech intensities. The gain factor is influenced by an estimation of the background noise level and the level of the loudspeaker signal. All environment parameters are adapted using near-end and far-end signals. For that purpose, a method of speech detection and a double-talk detector based on a cross-correlation estimate are introduced. The algorithm is implemented on a digital signal processor. Operation with a wide bandwidth of 12 kHz in various environments is satisfactory