Switch text between UPPERCASE, lowercase, Title Case, Sentence case, camelCase, PascalCase, snake_case, kebab-case and CONSTANT_CASE, with documented rules.
Prose
Keeps punctuation and line breaks
Identifier
Rebuilds each line as one token
Waiting for text
Each significant word is capitalised; a, the, of and other short joining words stay lowercase unless they are first or last. Words with an inner capital — iPhone, NASA — are left alone, unless the whole text is uppercase, which is treated as shouting and normalised.
Paste text to convert
Pick a style above and the result appears here as you type.
Case Converter runs entirely on your device — nothing is uploaded.
You’re on the Free plan. Compare plans
Nine styles, in two groups that behave differently on purpose.
The prose styles — UPPERCASE, lowercase, Title Case and Sentence case — keep the shape of your text. Punctuation, spacing and line breaks all survive; only the letters change. The identifier styles — camelCase, PascalCase, snake_case, kebab-case and CONSTANT_CASE — discard that shape by definition, because an identifier is a sequence of words joined by one separator. They tokenise each line and rebuild it, so a list of names stays a list rather than collapsing into a single identifier.
Title Case is where tools usually start guessing, so its rules are fixed and written down. Short joining words — a, an, and, of, the, to and a dozen more — stay lowercase unless they are first or last. A word that already has a capital after its first letter is left exactly as it is, so iPhone, NASA and McDonald survive. Hyphenated words get each part capitalised. Text that is entirely uppercase is treated as shouting and normalised, because that is the main thing people paste in.
Sentence case lowercases everything first and then capitalises the start of each sentence. That means acronyms and proper nouns are lost — which is what sentence case does. The tool does not try to protect them, because a rule that guessed which words to preserve would be wrong often enough to be worse than a rule you can predict.
There is no dictionary and no language model here. Every transformation is deterministic: the same input always produces the same output.
From a fixed list of short joining words — a, an, and, as, at, but, by, for, if, in, nor, of, on, or, per, the, to, via, vs and a few more. They are lowercased unless they are the first or last word. It is a deliberate, published list rather than grammatical analysis.
Because a word with a capital letter after its first character is treated as intentionally cased and left untouched. That protects iPhone, NASA, McDonald and product names generally. The exception is text that is entirely uppercase, which is read as shouting and normalised.
Sentence case lowercases everything and then capitalises sentence beginnings, so names and acronyms are lowercased with everything else. Distinguishing a name from an ordinary word needs a dictionary per language, and a tool that got it right most of the time would be harder to trust than one that is predictable.
Text is split on anything that is not a letter or a digit, and also at the boundaries inside existing identifiers: helloWorld becomes hello and World, HTTPServer becomes HTTP and Server, and digits form their own token. Accented and non-Latin letters are treated as letters.
It is discarded, because an identifier cannot contain it. Line breaks are kept, so each line becomes its own identifier. If a line has no letters or digits at all, it comes back empty and the tool says so.
No. Every conversion runs in your browser and nothing is sent, logged or stored.
