Yes, definitely. It adds complexity in that now you have another source of both Type I and Type II errors (failing to wake up, waking when it shouldn't). Voice ID itself is far from settled science to do well, so it would be a tradeoff.