In　order　to　synthesize　realistic　visual　speech　in　real　time,　this　paper　puts　forward　a　new　method　for　synthesizing　three-dimensional　facial　animation,　which　blends　dynamic　elements　of　visual　speech,　facial　expressions　and　other　related　factors.　Based　on　the　MPEG-4　facial　animation　standard,　two　databases　are　set　up　separately,　including　mouth　shapes　of　Chinese　syllable　and　basic　human　facial　expressions.　Then,　a　weighted　average　of　facial　animation　parameters　is　introduced　to　affect　facial　movement,　which　represents　variety　of　elements.　Experiment　results　show　that　the　method　proposed　in　this　paper　effectively　blends　multiple　elements　of　human　face,　receives　realistic　and　real-time　rendering　facial　animation.