The voice fingerprint — a client's voice, measured from their own copy — step 3 of 8

Windrow's live copy runs about this long. Past the ceiling, a variant stopped sounding like them — that's the number you'll defend to the client, not a regex. The house average is the mean of the three samples, rounded to 1 decimal; the ceiling is 1.5x that already-rounded number, rounded to the nearest int.

What does this print? (Two lines.)