Statistical language measures from speech · Anomaly
Counts and ratios taken from the transcript, without a language model in the loop.