System, apparatus, and method for improving speech quality in multi-party devices

DETAILED DESCRIPTION A method, system, and apparatus are provided, such that cross talk terms may be removed from speech signals in a multi-party conference system, having a plurality of users.
This is accomplished by first estimating the respective cross talk terms for each of the said plurality of users.
These terms are then used for controlling the transmission of speech and audio signals between the users.
In one approach, the transmission input speech signals are multiplied by time varying scalars, e.
g.
, taking the values 0 and 1 (on and off), thereby controlling the acoustic cross talk terms.
In another approach, line cross talk terms are inverted and multiplied to input speech signals, in effect filtering the input speech signals, such that respective source speech signals may be recovered from said plurality of multi-party conference users.
In the case of separation of sources, the signals are separate; thus original speech signals are not recovered.
Instead, signals are filtered versions and not identical to the corresponding source speech signals.