Live Word and Character Counter

1. 1

   Add your text

   Type or paste into the editor, or drop a TXT, MD, CSV, JSON, XML, HTML, SRT or VTT file. Nothing is uploaded — the file is read on your device.

2. 2

   Read the counts

   Words, characters with and without spaces, sentences, paragraphs, lines, unique words and syllables update as you type, alongside reading, speaking and handwriting time.

3. 3

   Set the rules that matter to you

   Change the reading speed, set a word-count goal or a character limit, and choose whether hyphens, apostrophes and numbers count inside a word.

4. 4

   Check density and readability

   The keyword panel ranks your commonest words and phrases with common words filtered out, and the readability panel gives Flesch Reading Ease and a Flesch–Kincaid grade for English prose.

5. 5

   Copy or save

   Copy any single count, copy the whole report, or download it as text — or as CSV or JSON on Pro, with a batch of documents saved as one archive.

Hit word limits, sanity-check length before submit, estimate read time, and copy a single number without a full document round-trip.

Benefits

• Limits: see totals against essay, ad, or platform caps.
• Speed: instant updates as you type or paste.
• Copy: one click per metric.
• Reading time: rough minutes from 225 wpm.
• Privacy: local-only processing.

Naive plain-text heuristics suitable for drafts—not a substitute for every publisher’s style guide.

Rules in this tool

• Words: whitespace-separated tokens that contain at least one letter or digit, so 3.14 and $5,000 are one word each.
• Characters: code points, so an emoji counts as one; a second count strips all whitespace.
• Sentences: split on . ! ? with an abbreviation table, so Dr. Smith and U.S.A. stay inside one sentence.
• Paragraphs: blocks separated by a blank line (double newline).
• Reading time: words divided by your chosen speed, 225 wpm by default, with a range for slower and faster readers.
• Quick stats: words per sentence and characters per word, shown as a dash rather than zero when there is nothing to divide.
• On-device: no network upload for the count itself.

How to interpret counts

• Use word count for assignment and article targets, and character count for platform-limited fields.
• Compare unique words and lexical diversity to detect repetition between revisions.
• Reading time is an estimate; use range values for conservative planning.

Most counters split on punctuation and accept whatever falls out. That is where counters disagree with each other and with Word, so the rules here are written down rather than left to a regular expression.

The decisions that change a number

• A number is one word. 3.14, $5,000 and 1,234.56 each count once. Splitting on the dot or the comma turns one figure into two or three words, which is the single most common way a counter inflates a total.
• An abbreviation does not end a sentence. Dr. Smith went home. He slept. is two sentences, not three. A title is followed by a name, so Dr. and Mrs. never end one; etc. and Inc. do end one when a capital follows.
• Initials stay together. J. R. R. Tolkien is part of its sentence, not four sentences. The trade is that a sentence genuinely ending in a single capital letter is joined to the next one — much the rarer case.
• A dotted acronym is one word. U.S.A. counts once and does not break the sentence when lowercase text follows it.
• A character is a code point. An emoji counts as one character, not two. That matters wherever a platform limit does.
• Scripts without spaces count per character. Chinese, Japanese, Thai, Lao, Khmer and Burmese have no spaces between words, so each character counts as one word — the convention every word processor and translation tool uses. Korean is space-separated and is counted normally.
• An unfinished last line still counts. A headline with no full stop is one sentence. Stricter sentence splitting requires the punctuation, which is what that toggle is for.

Three different jobs need three different rates, and a single "reading time" answers only one of them.

The rates used here

• Reading starts at 225 words per minute, within the 200–250 range usually cited for adult silent reading, and you can set it to anything from 80 to 600.
• The range beside it spans a quarter either side of your chosen speed, so it moves when you move the speed — a slower audience genuinely reads more slowly.
• Speaking uses 130 words per minute, the pace presentation coaches teach, which is what you want for a talk, a script or a voiceover.
• Handwriting uses 13 words per minute, for anyone planning an exam answer or a letter by hand.
• Pages use 250 words, the standard double-spaced manuscript page.

Density is how often a word or phrase appears as a share of the whole document. It is a drafting check, not a ranking factor — no search engine has published a target and none of them wants one.

Reading the panel

• Common words are hidden by default, because the ten commonest words of any English document are the, of, and, to and a.
• Two- and three-word phrases are counted separately, and no phrase crosses a full stop — so "the end. The next" never produces the phrase "end the".
• Use it to spot a word you have leaned on without noticing, and to compare two drafts of the same piece.
• Do not use it to hit a percentage. Writing to a density target is how copy stops sounding like a person wrote it.

Two scores from the same three inputs — words, sentences and syllables — reported only when there is enough English prose for them to mean anything.

What the two numbers say

• Flesch Reading Ease runs 0 to 100, and higher is easier. Roughly 60–70 is plain English; below 30 is heavy going.
• Flesch–Kincaid converts the same inputs into a US school grade level, which is the form most style guides quote.
• Both need at least 30 words. Below that a single clause swings the answer by whole grade levels, so no score is shown rather than a confident wrong one.
• Both are English formulas built on an English syllable estimate. They are not shown for text that is mostly Chinese, Japanese or Thai, where they would produce a precise and meaningless number.

Platforms do not count characters the same way, so one number cannot answer for all of them.

Where the counts differ

• X allows 280 and counts code points, so an emoji costs one.
• SMS allows 160 and counts UTF-16 units, so the same emoji costs two — and one non-GSM character drops the whole message to 70 per part.
• Instagram captions allow 2,200 and LinkedIn posts 3,000.
• A meta title is usually cut around 60 characters and a meta description around 160; both are pixel limits in practice, so treat them as guides.
• Set your own limit for anything else — a form field, a product title, a course abstract.

Paste is not the only way in. Drop a text file and the counter opens it on your device, works out its encoding and counts it.

What opens, and what does not

• Plain-text formats open directly: TXT, MD, CSV, TSV, LOG, JSON, XML, HTML, SRT, VTT, RST and TEX.
• UTF-8, UTF-16 and Windows-1252 are detected automatically, with or without a byte-order mark, so a file exported from an older editor still counts correctly.
• A byte-order mark is not counted as a character, because it is not one the author typed.
• DOCX, PDF, ODT and RTF are not plain text and are refused by name, with a link to the tool that extracts their text first.
• Pro counts up to a hundred documents at once and saves every report as one archive.

Blog drafts, homework, newsletters, social copy checks, and quick comparisons between two pasted versions.

Common use cases

• Academic: stay near assignment bounds.
• Web: headline and body length checks.
• Social: character-aware alongside word totals.
• Editing: spot empty splits (0 paragraphs) on single blocks.
• Reporting: paste figures into tickets or sheets.

Industry context is approximate; verify for binding requirements.

Key points

• Typical adult silent reading is often cited around 200–250 wpm; this tool starts at 225 and lets you change it.
• Academic targets and blog lengths vary widely by field and site.
• Sentence splits treat abbreviations and decimals naively—counts can inflate.
• A single block with no blank line is one paragraph; a blank line starts the next one.
• Counting runs beside the editor rather than inside it, so a large document never freezes typing.

Pair counts with your target spec.

Quality tips

• For strict limits, confirm the same rules as your editor or CMS.
• Use character count when the platform caps characters, not words.
• Re-count after cleaning HTML or Markdown to plain text.
• Reading time is a ballpark for typical prose.
• Compare drafts under the same paste hygiene.

When not to rely on it

• Legal, academic, or billing word counts that require certified tooling.
• Languages needing dictionary-aware tokenization.
• Exact paragraph rules that differ from blank-line detection.

Heuristic counts; requires JavaScript; works in modern devices.

Recommended workflow

• Run counter first to validate target length and draft structure.
• Use analyzer/statistics to investigate repetition, sentence complexity, and lexical spread.
• Finalize with readability and frequency tools before publishing.

For deeper review, pair this counter with text analyzer, text statistics, readability checker, and word frequency.

Powered by optimised processing algorithms.