I did wrestle a lot with that timing - it's probably just the animation throwing you off. The effect of pressing each button is instant, but the 'peak' of the actual animation is a couple of frames in. You'd have to have a keen eye for detail to notice but there you go! Perhaps something to consider in future animation designs for a UI like this.
That's a great idea about the frames showing something when they're pressed with no notes. This would also make it more clear for people less familiar with the DDR layout - which there have been a couple of feedback points on. Do you think that should also count as a miss and therefore cause the fish to drop further away? Or is it better just as an indication of which key you've pressed?