String Similarity Calculator

Compare two strings using five well-known string-distance and string-similarity algorithms at once: Levenshtein edit distance, Jaro and Jaro-Winkler similarity, Hamming distance, and cosine similarity over character bigrams. Where a plain text-diff tool shows you what changed, this tool tells you how much — as exact, reproducible numbers.

How to use it

Type or paste String A and String B, then read the results instantly — the tool recalculates on every keystroke. Turn on Ignore case to treat "Word" and "word" as equal across every metric.

What each metric means

• Levenshtein Distance — the minimum number of single-character insertions, deletions, and substitutions needed to turn String A into String B. The Levenshtein Similarity row expresses the same distance as a percentage (1 − distance ÷ longer string's length).
• Levenshtein Edit Operations — a breakdown of exactly how many insertions, deletions, and substitutions make up that minimal edit script, plus how many characters matched unchanged.
• Levenshtein Distance Matrix — the full dynamic-programming table the algorithm builds internally, row- and column-labeled with the actual characters compared. Read a cell as "the edit distance between the prefixes ending at this row and column." Skipped for strings longer than 25 characters, where the grid stops being readable.
• Jaro Similarity — rewards strings that share characters close together in position, tolerant of small transpositions (used in record linkage and duplicate-detection systems).
• Jaro-Winkler Similarity — Jaro similarity boosted for strings sharing a common prefix (up to 4 characters), which makes it a common choice for matching names and short identifiers where the start of the string tends to be more reliable.
• Hamming Distance — the number of differing characters at the same position, defined only when both strings are the same length; shown as "N/A" otherwise.
• Cosine Similarity (bigrams) — treats each string as a vector of overlapping 2-character sequences and measures the angle between them, a metric common in fuzzy text matching and near-duplicate detection.

Why do the algorithms disagree?

They're built for different jobs. Levenshtein and Hamming count raw character-level edits, so a single inserted letter near the start of a long string can look like a big change. Jaro-Winkler and bigram cosine similarity weigh shared structure more than position, so they're often more forgiving of typos, reordering, and length differences — which is why fuzzy-matching and spell-check systems tend to favor them over plain edit distance.

Is this the same as a text diff?

Not quite. Use this tool when you need a precise number — deduplicating records, tuning a fuzzy-matching threshold, grading how close a typed answer is to the correct one. For a side-by-side view of exactly which words or lines changed between two blocks of text, use the Text Compare tool instead, or the Text Similarity Checker for a general character/word-overlap score on longer passages.

Privacy

Everything runs in your browser — the strings you enter are never sent to a server.