there are different ways to do it, but one popular method is using blend shape. In this process, you duplicate the head and model it in your keyframe positions. Depending on how pefect you want your lip-synch, you could need quite a few duplicates. Each duplicate would have a different expression or vowel mouth form. Then, you select all your duplicates, hold shift, select your original and in your animation menu set, Deform > Create Blend Shape. Now, go to Window > Animation Editors > Blend Shape. You'll have a slider for each duplicate you created. As you raise the sliders, the original will morph into the selected duplicate expression. From there, it's just a matter of animating the sliders.