I am working on this project again. I was worried about keeping sound synchronized to video, but according to the following youtube video, sound can be up to 45msec early or up to 125msec late and still be perceived as simultaneous with the video. Because I am doing one frame every 130 msec, it looks like according to this information, I have a 170msec window of acceptable temporal variability for video framerate, in which I need to put the audio frames into.
Because some frames can be delayed, I should probably make the audio play nearly 45msec early, allowing it to get a full frame behind (130msec, with another 40msec to spare for extra long eink updates). This may take a bit of experimentation to find the optimal audio lead time, but should be pretty flexible.
EDIT: In that video, he says 125msec delay, but later says 100msec in the basketball demo. However, the shorter 145msec total delay still gives a full frame plus 15msec frame delay being perceptually accepted as simultaneous.