How much text is actually here?

What it counts

The Character Counter is a writing instrument, not a character-frequency analyser. It answers the practical questions: Will this fit in the field? How long will it take to read? How many words are in the draft? Is the paragraph count what I think it is? The numbers update as you type, so you can watch a caption shrink into a limit instead of pasting it into the page and guessing.

The main cards show characters, characters without spaces, words, unique words, sentences, paragraphs, reading time, and speaking time. The full report in the output pane adds letters, digits, punctuation, line counts, average word length, the longest word, and lexical diversity. That report is plain text, so it can be copied into a note or a ticket.

Reading and speaking time

Reading time is the word count divided by a chosen reading speed. The default is 200 words per minute, which is a reasonable average for adult prose. A dense technical document may be slower; a familiar email may be faster. The point is not precision but a sense of scale: a 1,200-word article is roughly six minutes at the default, and a 60-word caption is under twenty seconds.

Speaking time uses a slower default of 130 words per minute, which is closer to a measured presentation pace. A script that looks short on the page can take longer to say than expected, especially with pauses, slide changes, and the difference between reading and speaking. Use the speaking estimate for a talk or a voiceover, and the reading estimate for an article.

Word frequency and keywords

The two tables at the bottom are different views of the same text. “Most used words” counts every token, including the small grammatical words that hold a sentence together. It is the right table for spotting a repeated phrase, a tic, or a name that appears more often than you thought.

“Keywords” removes a built-in list of common English stopwords — articles, pronouns, prepositions, and auxiliary verbs — so the topic words rise to the top. It is a quick way to see what a draft is about, but it is not a search-engine keyword tool and it does not understand synonyms, phrases, or intent. Treat it as a mirror, not a strategy.

Examples

We shipped a small update today: faster search, clearer errors, and a fix for the export bug.

104 characters with spaces
89 characters without spaces
18 words
1 sentence

The exact count depends on whether the destination counts spaces, line breaks, and emoji. The tool shows both with-space and without-space counts so you can match the platform’s rule rather than guessing.

Today I want to show you three things: where the time goes, what we changed, and what happens next.

21 words
About 10 seconds at 130 wpm
1 sentence

At a measured pace, twenty-one words is a short opening — roughly ten seconds before pauses. The reading estimate would be faster, which is why the speaking speed is a separate choice.

The launch is ready. The launch page is ready. The launch email is not ready.

Most used: the (3), launch (3), ready (3)
Words: 15 · Unique: 9
Reading time: under 5 seconds

The repeated “launch” is not automatically a problem. It is a signal: the paragraph is about one thing, and the repetition may be intentional emphasis or a draft that has not been trimmed. The table shows the pattern; the editing decision is still yours.

Limits worth knowing

• Word counting is locale-aware. The browser’s word segmentation is used where available, which handles hyphenated words and apostrophes better than a naive split. Different languages have different ideas of what a word is, so a count is always an approximation.
• Sentence counting is heuristic. Abbreviations, decimals, and ellipses can confuse a sentence boundary. A sentence without terminal punctuation may be merged with the next one.
• Reading time is an estimate. It does not account for code blocks, tables, footnotes, or the difficulty of the prose. Use it for scale, not for scheduling.
• Keyword tables are English-biased. The stopword list is English. In another language, the “keywords” table may still contain common grammatical words, which makes it less useful but not wrong.
• Nothing is stored. The text disappears on refresh. Copy the report or the result before leaving.

Page updated 27 October 2026