Talk:Zero-crossing rate

What about signal values that are exactly zero?
This definition will fail to notice the zero crossings in signals that have a sample of exactly zero, e.g.,

s = [1, 0, -1, 0, 1, 0, -1, 0, 1]

Here there are obviously 4 zero crossings, but according to the given formal definition this signal would be said to have no zero crossings at all. In this case $$s_t s_{t-1} = 0$$ always, and is never less than zero.

Not sure how this corner case is handled in the literature, and not sure how to give a concise formal definition that avoids this problem. Maybe say "this definition applies only to signals that never contain the value zero", or else say "if your signal contains any zero samples, delete them from the sequence before applying this definition." — Preceding unsigned comment added by MusicScience (talk • contribs) 18:33, 29 October 2012 (UTC)