Audio words dont match lips after being on timeline

The raw clip off the sd card the auido is sycned perfectly with the person talking. Once pulled onto a timeline that matches the frame rates and hz sample of the clip its filmed in the words are two-three frames off. 

Attached is my sequence settings and my auido preferences settings. Changing the i/o buffer to 2048 has mad it almost sync up, any lower value and the audio is lagging behind by alot.