How We Measure Famous Voice Frequencies

What we measure

We estimate speaking fundamental frequency (F0) — the rate at which the vocal folds vibrate during ordinary speech, reported in hertz. The published figure is the median F0 across voiced frames, with a typical range spanning the 10th to 90th percentile of those frames. A median over voiced frames is robust to the two errors that dominate naive pitch measurement: octave glitches in the detector, and unvoiced stretches (consonants, breaths, silence) dragging the average.

Detection uses the McLeod Pitch Method with a per-frame clarity gate: a frame counts only if the detector is confident and the value falls in the plausible speech range (50–400 Hz). This is the same measurement our in-browser voice pitch test applies to your voice — so a comparison between your result and a published estimate is a like-for-like comparison, not two different instruments.

Which recordings qualify

• Natural speech only. Interviews, podcasts, unscripted conversation. Singing, shouting, and scripted dramatic performance are excluded — they use pitch differently than conversation does.
• No processed audio. Character voices with post-production (pitch shifting, layering, effects) describe an effect chain, not a person. Where a performed voice is the only thing people know, we grade the estimate C and say explicitly that it describes the performance.
• Multiple clips. Our target floor for a top confidence grade is three or more separate recordings and at least ninety seconds of voiced speech, sampled from different occasions.
• Recency matters. Speaking pitch drifts with age — typically downward for men into their 60s, and downward for women after middle age. Where a voice spans decades of recordings, we state which era the estimate reflects.

Confidence grades

What an estimate can't tell you

A median F0 is one number describing a living instrument. It moves with context — everyone speaks higher when excited and lower when relaxed — and with the recording chain: microphone distance, compression and broadcast processing all nudge the measured signal. Two careful measurements of the same person can legitimately differ by several hertz. That is why every figure we publish is worded as an estimate, carries a range, and carries a confidence grade.

Pitch is also not the whole of how a voice sounds. Resonance, pacing, articulation and texture do at least as much work as the fundamental — which is why two voices at the same Hz can sound nothing alike, and why our celebrity pages describe those qualities alongside the number.

Independence and rights

These estimates are produced independently by Deepr for informational purposes. We are not affiliated with, sponsored by, or endorsed by any person whose voice we describe. We do not host, redistribute or transform any recordings: source audio is measured, the source is cited, and the audio is discarded — only the numbers and citations are kept. Where a page shows a recording, it is embedded from the original YouTube channel and plays on YouTube's side. Portraits are free-license images (Wikimedia Commons or public domain) with their required attribution shown on the page, and we invent no quotes.