TOWARD SPEAKER INDEPENDENCE IN AUTOMATED LIP-SYNC