Why pace is not an axis
By Noam Shemla · Data & Method
TL;DR Nobody has ever been a good speaker because they spoke at 140 words per minute. Pace only matters through what it does to other qualities, so it is measured precisely, converted to a 0–100 value, and then split across Clarity, Tonality and Energy at different weights. It never appears on the radar.
Every measurement system faces the same temptation: you measured something precisely, therefore you score it. Words per minute is the easiest thing here to measure and the most precise number the product holds. Scoring it directly would have been the obvious move.
It is left unscored, and the reason is the most useful idea in the method.
Pace is a modifier, not a quality
Nobody admires or remembers a speaker for hitting 140 words per minute. What an audience notices is that a rushed delivery was hard to follow, or that a dragging one sounded like the speaker had lost interest in their own point.
So pace is kept as a measurement and dropped as a score, and the number is decomposed into the qualities it affects.
Deepgram nova-3, word-level timestamps with per-word confidence
Measured
- Overall words per minute
- The pace timeline
- The 0–100 pace value derived from it
Inferred, not measured
- Nothing. This conversion is arithmetic end to end; it simply never becomes an axis.
The curve
Two features of that shape are deliberate.
The plateau. A single ideal value would make the number twitch. Natural speech varies by ten or fifteen words a minute between takes of the same material, and no listener could detect the difference. A peaked function would still move your score every time, and a number that moves without a cause teaches you to ignore it. Thirty words per minute of flat ground means the value only changes when something changed.
The floors. An earlier version tapered to zero at 60 and 220. This one bottoms out at 40 and 50. Speaking far too fast makes you hard to follow, but still some way short of impossible to follow, and a function that reached zero would be asserting otherwise.
Where it goes
Clarity takes the largest share because pace damages comprehension first. Tonality takes a quarter because rate and vocal variety are entangled: monotone delivery tends to be evenly paced delivery. Energy takes the smallest share, because a fast talker often reads as animated whether or not they are.
This one-measurement-to-several-axes decomposition is the mechanism the entire Signature is built on. Pace is simply its clearest example.
Fast talkers and persuasion
The popular claim is that faster speech is more persuasive. Its origin is a 1976 study in which faster speakers were rated more knowledgeable and more persuasive than slower ones (Miller and colleagues).1
That finding did not survive intact. Later work found the effect depends on the listener's starting position and on how motivated they are to think: a faster rate helps when an audience is inclined to disagree, because it limits their ability to counter-argue, and can hurt when they are inclined to agree (Smith and Shaffer, 1995).2
What holds across studies is narrower, and it supports the design here rather than a target number. Extremes are penalised in both directions, and variation carries more than the average does, and variation is exactly what a single overall figure cannot see:
Limits of a words-per-minute figure
It does not know your language. The band was set against English and applied to everything. Languages differ substantially in how much meaning a word carries, so a syllable-dense language delivers the same content in fewer words and reads as slower.
It drops words it cannot spell. Tokens with no Latin letters are stripped before counting, so numbers and non-Latin scripts do not count. Number-heavy speech reads slower than it was.
It cannot see intent. A deliberate slow passage and a speaker who has lost the thread produce identical data.
Five bands ship at once The conversion function uses 130–160, and the prompt that writes your coaching agrees. But the pace coaching track says 120–150, the milestone that congratulates you uses 120–160, the resource recommender accepts 110–160 and tells you to aim for 130, and the gauge you watch while recording highlights 120–150. The conversion function is authoritative because it is the one that produces a number. The rest are coaching heuristics that were tuned separately and never reconciled.
Why speakers drift out of the band
Almost always, nerves. A rate above the band on material a speaker knows is compression under pressure. The fix is planned pauses at the transitions, which lower the rate as a side effect; "speak slower" is an instruction nobody has ever successfully followed mid-sentence.
Below the band, the usual cause is reading. Delivery from a script runs slow and flat because retrieval is replaced by decoding.
References
- Miller, N., Maruyama, G., Beaber, R. J., & Valone, K. (1976). Speed of speech and persuasion. Journal of Personality and Social Psychology, 34(4), 615–624. https://doi.org/10.1037/0022-3514.34.4.615
- Smith, S. M., & Shaffer, D. R. (1995). Speed of speech and persuasion: evidence for multiple effects. Personality and Social Psychology Bulletin, 21(10), 1051–1060. https://doi.org/10.1177/01461672952110006