nothing, not even close to.
the audio has 44100 or 48000 or even more "steps" per second in maximum. each of those represented in 32bits (or 16bits or what ever), and not "for two channels so 16bit each", but each channel on it's own, so stereosound even now has 64bits, a.k.a. 32bit left, and 32bit right.
and the steps are packets of those samples, of which you have 44100 per second (depends on the sampling rate). if you have 120bpm, you have 2bps. that means 22050 samples per beat. so you could slice it up to 22050 samples per beat, or, at a 4/4 rythm, 88200 steps possible..
and if you need even more steps, we obviously could set to a higher sample rate
so no, your 16 is wrong.