Communication Signal Analysis
See what
words hide.
Spot the signal.
RedFlag AI is a free body language and voice analyzer for YouTube videos. It tracks blink rate, gaze, brow, jaw tension, pitch, pauses and speaking rate — then flags the moments where the speaker's face and voice stop agreeing.
Live analyzer · Demo
Live demo: body language and voice signals, in real time.
A quick look at the analyzer in motion. Ready for the real thing? Paste a YouTube URL below — or open the full workspace.
Beta notice
This is the beta version of our solution. The more you use it, the smarter it becomes over time.
Currently, it supports only a single speaker in the frame. We're actively working to remove this limitation.
How it works
How to analyze a YouTube video in 5 steps.
- 01Paste the video link
- 02Allow RedFlag to access the tab so the video can load on the screen
- 03Press play
- 04When signals slightly diverge, the screen turns yellow; when the mismatch is strong, it turns red. That's your Red Flag — the rest is up to you.
- 05Download the processed RedFlagged video and share it with friends. Let them have fun too 🙂
About RedFlag
RedFlag is not a lie detector.
It is a cognitive assistant designed to help you notice moments where communication signals diverge.
There is no technology capable of objectively determining truth or falsehood. Human communication is complex, contextual, and deeply situational. Even the most advanced tools — from polygraphs to behavioral analysis — cannot replace human judgment.
What is possible, however, is identifying inconsistencies between different layers of communication. RedFlag analyzes verbal, paraverbal, and non-verbal signals during speech and highlights moments where these signals diverge. Such mismatches may indicate uncertainty, internal conflict, or low confidence in one's own words.
RedFlag does not draw conclusions for you. It simply surfaces "red flags" — moments that may warrant closer attention. The interpretation is always yours.
Science
What deception research actually shows.
Across psychology, linguistics, and behavioral science, studies consistently show that deception and manipulation are not expressed through a single universal signal. Instead, they emerge through patterns and contradictions across multiple channels of communication.
Nonverbal behavior
Facial expressions, micro-movements, eye activity.
DePaulo et al. (2003), Psychological Bulletin: no single cue is reliable on its own.
Paralinguistics
Tone, pitch, rhythm, pauses, speech rate.
Sporer & Schwandt (2006), Applied Cognitive Psychology: pitch and pauses shift under strain.
Cognitive load theory
Links mental strain to behavioral leakage.
Vrij, Fisher & Blank (2017), Legal & Criminological Psychology meta-analysis.
Confidence research
How internal belief shapes outward expression.
Bond & DePaulo (2006), Personality & Social Psychology Review.
There are no "sure signs" of lying — only probabilistic indicators. RedFlag follows this principle strictly. It highlights signal divergence, not "truth" or "lies."
Technology
13 body-language and voice signals RedFlag measures.
- Blink rate
- Gaze stability
- Head stability
- Brow furrow
- Jaw tension
- Smile authenticity
- Voice volume
- Pitch frequency
- Verbal stability
- Vocal activity
- Speaking rate
- Filler ratio
- Filler count
It detects deviations, identifies inconsistencies, and summarizes the data into simple, intuitive markers.
Ready to spot the signals?
Try RedFlag FreeFAQ
Frequently asked questions.
Is RedFlag a lie detector?
No. No technology can determine truth or falsehood, and RedFlag does not try. It measures a speaker's facial and vocal signals and flags the moments where they diverge from that speaker's own baseline. What a divergence means is always for you to judge.
What does RedFlag actually measure?
Thirteen signals: blink rate, gaze stability, head stability, brow furrow, jaw tension, smile authenticity, voice volume, pitch frequency, verbal stability, vocal activity, speaking rate, filler ratio and filler count. The red-flag verdict itself rides on three of them: pitch monotony, blink-rate elevation and voiced continuity.
How does RedFlag decide a moment is a red flag?
It compares the speaker to themselves. Blink rate is normalized against that speaker's own median across the whole video, pitch variation is compared with normal speaking ranges, and the fused deviation must persist for at least three seconds before a moment is flagged. Pausing, seeking or re-watching never changes a result. The full method, thresholds and test results are public: how RedFlag scores a video.
Which videos work best?
Any public YouTube video with one visible speaker: interviews, testimonies, podcasts, speeches. Podium and parliament footage works when the face fills at least a tenth of the frame height. Multiple speakers in frame are not supported yet.
Is RedFlag free?
Yes. Without an account you can analyze one video per day up to 10 minutes. A free account raises that to two videos per day up to 30 minutes each. Unlimited plans start at $9.99 for one month.
Where does the analysis run? Is my video stored?
The face tracking and voice analysis run in your browser, on your device, while the video plays. Analysis results may be kept for up to 30 days so you can revisit them; video and audio files are deleted once the analysis completes.
Can I use RedFlag for hiring, legal, security or medical decisions?
No. RedFlag signals are not evidence of deception, credibility, intent, health or personality, and the Terms of Service prohibit high-stakes use. It is a tool for noticing moments worth a second look, nothing more.
What research is RedFlag based on?
The multi-channel approach follows the finding that no single cue reliably indicates deception (DePaulo et al., 2003, Psychological Bulletin), the cognitive-load account of behavioral leakage (Vrij, Fisher & Blank, 2017), blink-rate research (Leal & Vrij, 2008), and the National Research Council's 2003 review of the polygraph. Plain-language explainers with full citations are in Learn.