Text tools · Counting
Indic word and character counter
Count the way a reader does: क्षि is one character, not four. Words, characters, code points, UTF-8 bytes, sentences (। ॥ .) and reading time, with a per-script breakdown.
- Words
- 0
- Characters
- 0what you see: क्षि counts as 1
- Code points
- 0Unicode scalars: क्षि = 4
- Bytes (UTF-8)
- 0size in a .txt file or database
- Sentences
- 0ends at । ॥ . ? !
- Reading time
- 0 secat 200 words/min
More counts
- Characters without spaces
- 0
- UTF-16 length (JavaScript, Excel)
- 0
- Lines
- 0
- Paragraphs
- 0
- Speaking time (at 130 words/min)
- 0 sec
Reading and speaking rates are assumptions you can change; people vary a lot.
How a basic counter sees this text
| Method | Count |
|---|---|
| Visible characters (this tool) | 0 |
| Code points: [...text].length | 0 |
| UTF-16 units: text.length, Excel LEN | 0 |
| Bytes: file size | 0 |
01 · Why counters disagree
One letter, four numbers.
An Indian-script letter is usually several Unicode code points: a consonant, a virama that joins it to the next consonant, and a vowel sign. A counter that counts code points, bytes or UTF-16 units counts the pieces. A reader counts the whole shape. The table is computed live by this tool’s own code.
| Text | Visible characters | Code points | UTF-16 units | UTF-8 bytes |
|---|---|---|---|---|
| क्षिone letter-with-vowel-sign (क + ् + ष + ि) | 1 | 4 | 4 | 12 |
| हिन्दीa six-letter word | 2 | 6 | 6 | 18 |
| भारतthree aksharas | 3 | 4 | 4 | 12 |
| ಕನ್ನಡKannada conjunct ನ್ನ | 3 | 5 | 5 | 15 |
| স্ত্রীBengali conjunct স্ত্র + ী | 1 | 6 | 6 | 18 |
| தமிழ்Tamil, pulli on the last letter | 3 | 5 | 5 | 15 |
| मराठी ✌🏽with an emoji and skin tone | 5 | 8 | 9 | 23 |
02 · How it counts
Four decisions, spelled out.
- 01
Characters = grapheme clusters
The browser’s text segmenter groups a base letter with its vowel signs and marks. We then join virama + consonant, which the standard covers only for some scripts, so Kannada ಕ್ಕ and Tamil-adjacent cases behave consistently. Tamil’s pulli stays separate because த் is a letter in its own right.
- 02
Zero-width joiners stay attached
ZWJ and ZWNJ, used to force half-forms in Devanagari and to fix Malayalam chillus, join the cluster they belong to. They add code points and bytes but never characters.
- 03
Words are space-separated tokens
A token counts as a word when it contains a letter, vowel sign or digit. Hyphenated compounds count once.
- 04
Sentences end at । ॥ . ? ! …
Followed by a space, a closing quote or bracket, or the end. Decimal points and dots inside web addresses are ignored.
03 · FAQ
Questions about indic word and character counter
How many characters is “क्षि”?
One, as a reader sees it, but four to a computer: क + ् (virama) + ष + ि. A tool that counts Unicode code points says 4, one that counts UTF-16 units (JavaScript’s .length, Excel’s LEN) says 4, and a byte counter says 12, since each Devanagari code point takes 3 bytes in UTF-8. This tool counts the visible character, 1.
Why do Word, Google Docs and other counters give different numbers?
Word processors have their own rules for what counts as a character, and they differ between products and versions for Indian scripts. Some count code points, some count grapheme clusters. For anything with a character limit, such as an SMS, a form field or an exam answer, check which count the receiving system uses.
How are words counted?
Words are runs of text separated by spaces, tabs or line breaks that contain at least one letter, vowel sign or digit. A dash or a stray danda on its own is not a word. Hindi, Tamil, Bengali and the other Indian scripts separate words with spaces, as English does, so this works the same way.
What counts as a sentence?
Text ending in a danda (।), double danda (॥), full stop, question mark, exclamation mark or ellipsis, followed by a space or the end of the text. A dot inside a number or a web address (3.14, a.b.com) doesn’t end a sentence.
How is the reading time worked out?
Words divided by a reading speed. The page assumes 200 words per minute for silent reading and 130 for reading aloud; both are rough assumptions and you can change the first. People vary a lot, and Indian-language texts with long compound words read more slowly.
Does it work for text that mixes English and Hindi?
Yes. The “By script” panel shows how many characters are Devanagari, Latin, digits, punctuation or spaces.
Is my text uploaded?
No. Counting runs in your browser and nothing is sent anywhere.