In g729 decoder process, it decode bitstream to constrcut excitation,then pass

LP synthesis filter,and get synthesized speech. but it omit long-term (pitch synthesis)

filter before LP synthesis filter. Why? what's wrong? thanks!