Does your voice change under stress? Pitch, monotony and the evidence

← All articles · By the RedFlag team · September 4, 2026

The voice is the best-studied channel in deception and stress research — and the one with the most oversold commercial products. Two findings hold up: pitch rises slightly under stress, and pitch variability collapses under cognitive strain. Everything beyond that, treat with suspicion.

What replicates

Classic experiments found small but real elevations in fundamental frequency (F0 — perceived pitch) when speakers were under stress or lying (Streeter et al., 1977; Ekman, Friesen & Scherer, 1976). Sporer and Schwandt's (2006) meta-analysis of paraverbal cues confirmed pitch as one of the few acoustic variables with a consistent, if modest, association with deception. The effect is a fraction of a semitone — real in aggregate, useless as a per-moment lie test.

The second finding is about range, not height. Engaged, comfortable speech is melodic: pitch moves through roughly 2.3–4.5 semitones of variation. Under high cognitive load, speech flattens toward monotone — the speaker's planning resources are occupied and prosody is what gets sacrificed. Speech-based cognitive-load measurement research treats reduced pitch variability and changed pause patterns as core features precisely because they replicate.

Where the field went wrong: "voice stress analysis"

Commercial voice-stress analyzers have been sold to police departments for decades, claiming to detect deception from "microtremors" in the voice. Independently tested, they fail. Harnsberger and colleagues (2009) evaluated layered voice analysis under controlled conditions: performance at chance. A field study funded by the U.S. National Institute of Justice (Damphousse, 2007) found VSA detected about 15% of lies about drug use — while arrestees who believed the machine worked lied less, which is a deterrence effect, not detection. If you remember one thing about voice-based deception claims, remember that number.

Monotony is a signal you can hear yourself

Flattened pitch is audible: answers that go flat and stay flat, especially from a speaker who is melodic elsewhere in the same conversation, are moments of unusual effort or control. That is not a lie indicator. It is a difference indicator — deliberate control, rehearsed delivery, fatigue, and difficult recall all flatten prosody too. The honest use of the signal is to ask "what is happening here?", not to conclude.

How RedFlag uses pitch

Pitch monotony is the anchor signal of RedFlag's verdict, at weight 0.75: pitch variability (semitone standard deviation over a centered ±5-second window) is compared against the normal 2.3–4.5-semitone speaking band, and sustained flattening below it raises the deviation. There is deliberately no "stress from microtremors" anywhere in the pipeline. RedFlag is not a lie detector; the thresholds and test results are public: how RedFlag scores a video.

References

• Streeter, L. A., Krauss, R. M., Geller, V., Olson, C., & Apple, W. (1977). Pitch changes during attempted deception. Journal of Personality and Social Psychology, 35(5), 345–350.
• Ekman, P., Friesen, W. V., & Scherer, K. R. (1976). Body movement and voice pitch in deceptive interaction. Semiotica, 16(1), 23–27.
• Sporer, S. L., & Schwandt, B. (2006). Paraverbal indicators of deception: a meta-analytic synthesis. Applied Cognitive Psychology, 20(4), 421–446.
• Harnsberger, J. D., Hollien, H., Martin, C. A., & Hollien, K. A. (2009). Stress and deception in speech: evaluating layered voice analysis. Journal of Forensic Sciences, 54(3), 642–650.
• Damphousse, K. R. (2007). Voice stress analysis: only 15 percent of lies about drug use detected in field test. NIJ Journal, 259.