This is a frequent topic on the COW, so a search will probably yield the answer you want. Basically, you want to use “Convert audio to Keyframes” on your audio layer, creating a new layer of keyframe values based on the audio amplitudes. Then use a position expression on your jaw layer that will move it up and down according to the keyframes of this new audio keyframes layer. I have successfully done it a few times by reading other more experienced users explanations of the technique, so you may want to read further.