Body Language and Speech Analytics for the Detection of Deception
It is well known that humans are highly tuned to body language. More than half of communication happens through visual cues generated by the human body. The way we move, we sit, the positioning of our hands, movement of our head, all convey information about the states of our minds to others. Similarly, non-verbal vocalizations convey a tremendous amount of information about the state of the speakers. Non-verbal cues such as pauses, energy level, intonations, etc., convey the emotional state of the speaker to the listener. When humans combine body language observations with non-verbal cues, they develop a very nuanced picture of the person / persons with whom they are communicating without regards to the actual subject of the conversation.
The Kovid Thinking Machine recognizes and analyzes multi-modal signals such as body language cues and non-verbal vocalizations in a mechanized manner. This analysis can occur under multiple scenarios, with the scenario determining the context and thereby the qualitative assessment.
- In cases where the interactions are between two parties which lasts several minutes, the KTM produces a subjective output based on the overall pattern of the interaction.
Applications: Banks and other financial institutions can use this capability to help assess borrower risk when dealing with high value loans
- In cases where only visual observations are needed, the KTM can act on just the body language of individuals, or groups, and produce subjective assessments.
Applications: There are a wide variety of applications for this capability:
- Businesses dealing with high value customers can sense if their customers are getting agitated when waiting for services and have a senior employee intervene
Correctional Facilities would be able to leverage this capability to monitor the offenders and discern violence or tense confrontations that could spiral out of control