Here's what I would do (I've never actually tried it yet, but I imagine it will work):

Go to a certain spot in the audio where you want a particular movement to start. Copy down what the timecode says. Go to the spot where you want the movement to end. Copy down that timecode. Subtract the two numbers. Divide that number by your framerate. The number you get is the number of frames the synced movement will be.
(This also works for figuring out the time between movements.)

Just do what I told you.

If you want a specific example (almost wrote exaMLPe): What you want synced is the piano player playing the piano to the music. The spot (almost wrote spit) the note is heard is when you want the player's hand on a key. This is the end of the movement. The previous note is the beginning of the movement, when the hand is on a different key. Do my method, and you should get the number of frames it will take for the player to lift off one key and press another.

Figuring out how many notes are in a second won't work; the notes being irregular. But it would work if you want to sync a movement to a regular beat: find out the amount of frames from one beat to the next (using my method) and you'll know how many frames are between each beat.

Also, never use computer pans. In-shot pans look so much better because of perspective; If a camera (or your eye, for that matter) is moving across a space, the closer objects will appear to move by faster than distant objects. A computer-generated pan can't replicate this, which is why it doesn't look right.

You COULD download Audacity, and then accurately select pieces of the song to get a precise measure.

Also dude I am totally looking forward to this video that's a really pretty song that you've chosen to make a brickfilm so good luck