This tells what people mean, not just what they say. If they’re uncomfortable, it gets flagged. If rate of speech deviates, it is flagged. If they are off script, it is flagged.

The dataset has the transcripts and audio of the earnings call, along with comprehensive data indicating higher / lower areas of stress by call time and topic, in addition to increases / decreases in rate of speech and vocabulary changes.

The dataset provides a clean signal of deltas. If lawyers start writing the introductory text to the call, the word patterns will flag it. If different adjectives are used to describe earnings, it is flagged. Because each executive is profiled individually, unique vocal quirks and stress patterns highlight when a person is more / less comfortable discussing a particular topic, which when trended, can highlight areas of exploration or risk.