- Reading time
- 0 seconds
- Read aloud
- 0 seconds
- Average word length
- 0
- UTF-8 bytes
- 0
- UTF-16 units (String.length)
- 0
- Unicode code points
- 0
How it works
Counting characters seems trivial until an emoji shows up. The number a person sees does not match what a database stores or what a programming language returns: the family emoji is a single character on screen, but it is 5 Unicode code points, 8 UTF-16 units (what String.length returns in JavaScript, Java or C#) and 18 bytes in UTF-8. This tool shows all four measures, so it works both for marketing copy and for sizing a VARCHAR column.
Words, sentences and characters are counted with the Unicode segmentation rules built into the browser, the same ones it uses to move the cursor or select a word with a double click. That is why it also works for languages that do not separate words with spaces, such as Chinese or Japanese, and does not count a letter with a combining accent as two characters.
Reading time uses the average speeds from a meta-analysis of 190 studies (Brysbaert, 2019): 238 words per minute for silent reading of non-fiction and 183 when reading aloud. The text never leaves your browser and is not stored in the share link.
Examples
👨👩👧1 · 5 · 8 · 18One visible character, 5 code points (three emojis joined by two invisible joiner characters), 8 UTF-16 units and 18 bytes in UTF-8. If an API limits you to 280 "characters", it is worth finding out which of these measures it uses.SMS · 151 GSM-7 → 151 UCS-21 SMS → 3 SMSA 151-character SMS fits in a single message with the GSM alphabet, but a single í is enough to make the phone switch to UCS-2, where each part holds 67 characters: the same text now costs three messages.café (NFC) · cafe + ◌́ (NFD)4 · 5The same word can be stored with é as a single code point or as an e followed by a combining accent. They look identical on screen and this tool counts 4 characters in both cases, but String.length returns 4 or 5.Use cases
- Fit titles and meta descriptions to the length Google shows without cutting them off.
- Work out how many SMS a campaign or a verification code will cost, and spot the character that makes it more expensive.
- Size a database column or validate a field with a byte limit, such as VARCHAR in MySQL with utf8mb4.
- Meet the minimum or maximum word count of an assignment, an abstract or a call for papers.
- Estimate how long an article takes to read or a script takes to record.
- Spot overused words before publishing a text.
Frequently asked questions
Why does another counter give me a different number?
Because there is no single way to count. Many counters use String.length and count UTF-16 units, so an emoji counts as 2 or more; others split words with regular expressions that break at apostrophes and count "don't" or "l'eau" as two. Here characters are what you see on screen and words follow the Unicode rules, which is the closest thing to how a person counts.
Why does an accent change the cost of an SMS?
The 7-bit GSM alphabet allows 160 characters per message but does not include every accented letter: it has é, ñ, ü and à, but not á, í, ó or ú. With a single character outside that alphabet, the whole message switches to UCS-2 and the limit drops to 70, or 67 per part if it has to be split. Some carriers use the Spanish national table, which does include them, but it is best not to count on it.
How many characters can a title for Google have?
Google does not cut by characters but by width in pixels, around 600 on desktop, so the real limit depends on the letters: a W takes up much more than an i. Between 50 and 60 characters usually fits in full; the meta description is shown up to about 155 to 160 characters. The tool's bars use those values as a reference.
Does the count match X (Twitter)?
Almost always, but not entirely. X counts every link as 23 characters regardless of its length and Chinese, Japanese and Korean characters as 2. For texts without links in the Latin alphabet, the number is the same.
How many bytes does a text take up in the database?
It depends on the column's encoding. In UTF-8, English letters take 1 byte, accented letters and Cyrillic 2, Chinese 3 and emojis 4. MySQL with utf8mb4 counts VARCHAR length in characters but the row limit in bytes, and SQL Server with NVARCHAR uses UTF-16, which is this tool's UTF-16 units measure.