A software program has been designed that successfully runs a
stereophonic acoustic echo canceler natively on a personal
computer. This is a major achievement since an echo canceler
requires that the soundcard's input and output signals are
time-synchronous. Synchronizing the audio streams is a great
challenge in such an ``asynchronous'' environment as the operating
system of a PC. Furthermore, stereophonic echo cancellation is
significantly more complicated to handle than the monophonic case
because of computational complexity, nonuniqueness, and
convergence problems. Special care has to be taken in the
algorithm design. This work presents the core algorithms, i.e.,
the adaptive algorithm and the double-talk detection algorithm of
the implemented echo canceler.
