Open Access Open Access  Restricted Access Subscription Access

Comparative Analysis of a Lie Detector using Support Vector Machine, Naive Bayes, and Random Forest Algorithm with Speech to Text Conversion

Harshada Desai, Sanket Chavan, Neha Deshmukh, Abhijeet More


Lie detection, additionally known as deception detection, uses questioning techniques to determine truth and falsehood in response. Physiological responses like vital sign, blood pressure, heartbeat, and respiratory rate are used to discriminate between truth and lie. Once we lie, our blood pressure goes up, our heart beats quicker, we have a tendency to breathe faster (and our breathing slows once the lie has been told), and changes occur in our skin moisture. The recorded changes in these parameters are corrected and analysed in relevance specific queries. It is discovered that physiological options i.e., voice pitch, skin conductance, and heart rate variability are correlative to the range of high-stress things that helps to seek out deceptive and truthful behaviour on completely different topics. In this paper, features of speech and physical values are used from a dataset to determine truth and lie. The obtained results are primarily based upon comparative analysis of Random Forest, Naïve Bayes, and Support Vector Machine (SVM). We have implemented a real-timelie detector that uses certain parameters and produced a result.


Random Forest, Naïve Bayes, SVM (support vector machine), speech to text conversion

Full Text:




  • There are currently no refbacks.

Copyright (c) 2021 Research & Reviews: A Journal of Embedded System & Applications