Removing reverberative echo components in speech signals

摘要

A system is disclosed for automatically reducing reverberation in typical voice telecommunications systems. This system uses center clipping levels adaptive to the level of reverberation input speech. In one configuration, the voiceband is divided into two sub-bands and center clipping occurs only for the lower band; costs are reduced with surprisingly little transmission quality sacrifice. Clipping-level holdover circuitry with exponential decay appears to work well for a large variety of reverberant enclosures.