Estimate how long text takes to read silently or say aloud — handy for articles, scripts and speeches.
Paste text (or type a word count below):
Average adult silent reading speed is around 238 words per minute; speaking aloud (presentations, voiceovers) is slower, around 130 wpm. Multiply word count by the appropriate rate. Actual speed varies with difficulty and audience, so treat these as solid ballparks.
The 238 wpm figure comes from a 2019 meta-analysis of reading-rate studies, which put silent reading of English non-fiction at roughly that speed for adults. It is an average across many readers and many texts, not a target.
| Material | Typical pace |
|---|---|
| Fiction, light prose | Faster than average — familiar vocabulary, predictable structure |
| Non-fiction articles | Around the 238 wpm average |
| Technical documentation, code | Considerably slower; readers stop, re-read and cross-reference |
| Read aloud, presenting | Around 130 wpm, and slower still with pauses for emphasis |
Anything with tables, equations or code samples breaks the word-count model entirely, because time goes into parsing structure rather than consuming words in sequence.
Working backwards from a duration is more reliable than trimming afterwards. At 130 wpm, a five-minute talk is about 650 words and a ten-minute conference slot about 1,300. Those numbers assume you say every word at a steady pace, so budget below the limit: pauses, a demo that stalls, or a question mid-flow will eat the margin.
For video scripts, count only spoken narration. On-screen text, silent demonstrations and transitions add runtime without adding words, and a script that times perfectly on paper regularly overruns on camera.
They assume different rates. Anywhere from 200 to 275 wpm is in common use, and some platforms add a fixed number of seconds per image. On a 1,000-word post the choice of assumption moves the estimate by more than a minute, which is why two sites can label the same article differently.
Headings and captions, yes — they are read. Code blocks are the awkward case: readers either skip them entirely or spend far longer than the word count implies. If a post is mostly code, the reading-time estimate says very little about how long it actually takes.
Drafting with AI? Pair with the Token Counter.