Samples
On this page, you can find samples demonstrating the capabilities of voice-based deception detection technology. The reports presented below for each video were built in a fully automatic mode by analyzing the audio track in the video file.
Testing Deception Detection with "Box of Lies" Segment
In this part of The Tonight Show Starring Jimmy Fallon, guests participate in a game called "Box of Lies," where they take turns describing unusual objects hidden in boxes. The objective is to either truthfully describe the object or make up a lie, while the opponent tries to guess whether the description is true or false based on voice and behavioral cues.
This setup provides an ideal testing scenario for our voice-based deception detection system. The spontaneous, conversational nature of the game, combined with the intentional attempts at deception, offers a valuable dataset to analyze subtle changes in speech patterns, tone, and vocal cues. By processing these recordings, the system can be trained and tested to differentiate between truth-telling and deceptive speech.
A game show is one of the most challenging environments for analysis, as background music and audience cheers often make voice recognition more difficult. Additionally, professional actors participate in the game, and they are skilled at hiding their true emotions and telling convincing lies. Despite these challenges, our model is capable of detecting deception even under such conditions, which is a significant achievement in the field of automated deception detection.
Below, you can find two examples of the "Box of Lies" segment with different guests. Each video is accompanied by a report generated by our system, which highlights the key indicators of deception detected in the guest's speech. The reports provide a detailed analysis of the vocal features, linguistic patterns, and behavioral cues that contribute to the detection of deception.
In this example, you may encounter such a phenomenon known as "lying in a TV show". What does it mean? You may notice that when the host greets the participant and the audience at the beginning of the game, it is recognized as a lie, although it seems that the host has no direct motive. However, think about this. Is he really happy to greet the participant and the audience? Such shows are filmed several times a day and there are unsuccessful takes, for him it is just hard work, so it is quite natural to be unhappy to greet the audience, which is what the deception detector algorithm detects.
You also most likely cannot know the motive for telling a lie or the truth of the participants of the show when it is not about describing objects in a box. Therefore, in order to evaluate the quality of the algorithm of the deception detector, you should, in this show, pay attention to those moments when the participant or the host describes the object, then you will surely know whether he is telling the truth or lying, and compare with what the detector has determined and is shown in the report.
Other samples coming soon...
Testing Deception Detection with Youtube Chanell Academy of Study Lie
The video on this page was provided by Youtube channel of the Academy of Lie Studies. Each video is an interview where the participant is asked questions to which they answer as they see fit, they can tell the truth, or they can lie. In the second part of the video, which is presented below the report, the participant tells where he lied, which makes it possible to check the reliability of the data in the lie detector report.
Please note that the reports in this category are subject to the restrictions of the tariff plan specified in the report title. This means that some reports may be time-limited and may not cover the entire video clip in time. This done for demonstrate the difference in tariff plans.
This show has a drawback. The participants do not reveal the whole truth. Sometimes they directly indicate that here there was a lie, and as for the rest, participant will not discuss where the truth is and where the lie is. Therefore, pay attention when reading the report to those moments when the lie is revealed by the participant.
Other samples coming soon...
Under construction