What is speaking rate?
By Noam Shemla · Measurements
Speaking rate is how many words you produce in a minute of talking. It is the simplest measurement in speech analysis and the one most often reported without saying what was divided by what.
The short version We divide your word count by the full length of the recording, including every silence. Most quoted rates divide by speaking time only. Ours will therefore read lower than a figure you find elsewhere for the same performance. That is deliberate: pauses are part of delivery.
Dividing words by time
The recording is transcribed with word-level timestamps. Words are counted after stripping punctuation and anything that is not a letter, then divided by the clip's full duration, as the recorder measured it.
Deepgram nova-3, word-level timestamps with per-word confidence
Measured
- Overall words per minute
- Words per minute per 10-second bucket
Inferred, not measured
- Nothing. Pace is converted to a 0–100 value by a published function and then split across Clarity, Tonality and Energy, and is never scored as an axis of its own.
There are two rates in your report and they answer different questions. The overall figure includes every silence, so it describes the density of a performance. The timeline counts the words that start in each 10-second slice and multiplies by six. A slice with no words at all produces no point rather than a zero, so a long silence pulls the overall figure down further than it pulls the timeline, which comes closer to how fast you were going while you were talking.
A recording that falls back to the estimated path, because the transcription came back with low confidence, has no word timestamps. Its rate is still word count over the full duration, taken from the transcript, but it has no timeline.
What rate does to a listener
Rate carries information the words do not. Speech that is too fast reads as nervous and gives an audience no room to process; too slow reads as unsure, and listeners fill the gap with their own thoughts.
The research is more equivocal than the advice usually given. An often-cited early finding is that faster speakers were rated as more knowledgeable and persuasive than slower ones (Miller and colleagues, 1976).1 Look at their second experiment closely and most of the rise in agreement came between a slow recording at 111 words a minute and a moderate one at 140; going on to 191 added little. The damage was done by being slow, which is a narrower lesson than "talk fast".
Later work qualified the finding substantially. Smith and Shaffer found that fast speech made a message listeners disagreed with more persuasive, by leaving them less room to argue back, and a message they agreed with less persuasive, by leaving them less room to think of reasons to agree (1991).2 In a follow-up, fast speech blurred the difference between strong and weak arguments, and raised persuasion only when the topic mattered moderately to listeners, through the speaker's credibility (1995).3
What survives across studies is narrower and more useful: extremes are penalised, and variation matters more than the average. A rate that never changes is the problem, not the rate itself.
The pace band
- 130–160 full marks. The plateau where our pace score is 100.
- ~150 conversational english. The commonly cited rate for relaxed speech.
- 115 / 175 where the fall stops. At 115 words a minute or below the score is 50; at 175 or above it is 40. It never reaches zero.
The band our score rewards, against the rate ordinary conversation runs at.
Between the plateau and the floors the score falls in a straight line. Two worked examples: at 168 words a minute you are eight words into a fifteen-word fall that loses 60 points, so the pace score is 68. At 122 you are eight words into a fifteen-word fall that loses 50, so it is about 73. The fall is steeper on the fast side and the floor lower. The floors are there because even a badly mismatched pace leaves a speaker partly intelligible.
Because most people already fall inside the band in ordinary speech, a low pace score usually means something specific happened: nerves compressing the delivery, or reading from a script.
Where the pace score goes
Pace is never scored as its own axis. The 0 to 100 value is fed into three of the five: Clarity at a weight of 0.6, Tonality at 0.25 and Energy at 0.15. Each axis also has the coaching model's own score at a weight of 1.0, so pace corroborates those judgements rather than replacing them. The reasoning is in why pace is not an axis.
Blind spots of a words-per-minute figure
It does not distinguish a fast, clear speaker from a fast, unintelligible one. Rate is words per unit time; it says nothing about articulation.
It does not know your language. The rate is computed identically whatever you speak, and the Signature's pace band is the same for every language, but languages differ in how much a syllable or a word carries. A study of seventeen languages found that those with dense syllables are spoken more slowly and those with light syllables faster, so that they convey information at a broadly similar rate, about 39 bits a second.4 Words per minute does not see any of that. The 130 to 160 band was set against English and is applied to everything.
And it does not know what you were doing with the silence. A speaker who pauses deliberately three times in a minute and one who forgets their line three times produce the same overall rate.
Using your own number
Read your overall rate once, then spend your attention on the timeline. Look for the slices where you sped up, and check what you were saying there. Speeding through your main point is the common pattern, and slowing on purpose for that one sentence is a better change than moving your average.
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. (1991). Celerity and cajolery: rapid speech may promote or inhibit persuasion through its impact on message elaboration. Personality and Social Psychology Bulletin, 17(6), 663–669. https://doi.org/10.1177/0146167291176009
- 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
- Coupé, C., Oh, Y. M., Dediu, D., & Pellegrino, F. (2019). Different languages, similar encoding efficiency: Comparable information rates across the human communicative niche. Science Advances, 5(9). https://doi.org/10.1126/sciadv.aaw2594